feat(rate): add synchronized flight sample inspector

.github/workflows/rate-web-playwright-trusted.yml

@@ -8,6 +8,8 @@ on:
       - "src/rate_of_closure/club/**"
       - "src/rate_of_closure/club_camera.py"
       - "src/rate_of_closure/club_mesh_source.py"
+      - "src/rate_of_closure/flight_accepted_study.py"
+      - "src/rate_of_closure/flight_sample_inspector.py"
       - "src/rate_of_closure/mesh.py"
       - "src/rate_of_closure/model.py"
       - "src/rate_of_closure/plotting/**"
@@ -24,11 +26,18 @@ on:
       - "src/shared/python/swing_sim/putting.py"
       - "tests/rate_of_closure/pyqt_putting_sample_inspector_probe.py"
       - "tests/rate_of_closure/pyqt_club_camera_probe.py"
+      - "tests/rate_of_closure/pyqt_flight_sample_inspector_probe.py"
       - "tests/rate_of_closure/test_club_camera.py"
       - "tests/rate_of_closure/test_club_mesh_source.py"
       - "tests/rate_of_closure/test_club_view_camera.py"
+      - "tests/rate_of_closure/test_flight_accepted_study.py"
+      - "tests/rate_of_closure/test_flight_explorer.py"
+      - "tests/rate_of_closure/test_flight_explorer_atomic_gui.py"
+      - "tests/rate_of_closure/test_flight_sample_inspector.py"
+      - "tests/rate_of_closure/test_flight_sample_inspector_gui.py"
       - "tests/rate_of_closure/test_mesh.py"
       - "tests/rate_of_closure/test_pyqt_club_camera_rendered.py"
+      - "tests/rate_of_closure/test_pyqt_flight_sample_inspector_rendered.py"
       - "tests/rate_of_closure/test_pyqt_putting_sample_inspector_rendered.py"
       - "tests/rate_of_closure/pyqt_variation_render_probe.py"
       - "tests/rate_of_closure/test_pyqt_variation_rendered_interactions.py"
@@ -91,6 +100,7 @@ jobs:
           tests/rate_of_closure/test_pyqt_variation_visual_state_rendered.py
           tests/rate_of_closure/test_pyqt_putting_sample_inspector_rendered.py
           tests/rate_of_closure/test_pyqt_club_camera_rendered.py
+          tests/rate_of_closure/test_pyqt_flight_sample_inspector_rendered.py
           tests/rate_of_closure/test_pyqt_visualization_tab_visibility.py -q -n 0
 
       - name: Retain Playwright evidence

.github/workflows/rate-web-playwright.yml

@@ -7,6 +7,8 @@ on:
       - "src/rate_of_closure/club/**"
       - "src/rate_of_closure/club_camera.py"
       - "src/rate_of_closure/club_mesh_source.py"
+      - "src/rate_of_closure/flight_accepted_study.py"
+      - "src/rate_of_closure/flight_sample_inspector.py"
       - "src/rate_of_closure/mesh.py"
       - "src/rate_of_closure/model.py"
       - "src/rate_of_closure/plotting/**"
@@ -23,11 +25,18 @@ on:
       - "src/shared/python/swing_sim/putting.py"
       - "tests/rate_of_closure/pyqt_putting_sample_inspector_probe.py"
       - "tests/rate_of_closure/pyqt_club_camera_probe.py"
+      - "tests/rate_of_closure/pyqt_flight_sample_inspector_probe.py"
       - "tests/rate_of_closure/test_club_camera.py"
       - "tests/rate_of_closure/test_club_mesh_source.py"
       - "tests/rate_of_closure/test_club_view_camera.py"
+      - "tests/rate_of_closure/test_flight_accepted_study.py"
+      - "tests/rate_of_closure/test_flight_explorer.py"
+      - "tests/rate_of_closure/test_flight_explorer_atomic_gui.py"
+      - "tests/rate_of_closure/test_flight_sample_inspector.py"
+      - "tests/rate_of_closure/test_flight_sample_inspector_gui.py"
       - "tests/rate_of_closure/test_mesh.py"
       - "tests/rate_of_closure/test_pyqt_club_camera_rendered.py"
+      - "tests/rate_of_closure/test_pyqt_flight_sample_inspector_rendered.py"
       - "tests/rate_of_closure/test_pyqt_putting_sample_inspector_rendered.py"
       - "tests/rate_of_closure/pyqt_variation_render_probe.py"
       - "tests/rate_of_closure/test_pyqt_variation_rendered_interactions.py"
@@ -96,6 +105,7 @@ jobs:
           tests/rate_of_closure/test_pyqt_variation_visual_state_rendered.py
           tests/rate_of_closure/test_pyqt_putting_sample_inspector_rendered.py
           tests/rate_of_closure/test_pyqt_club_camera_rendered.py
+          tests/rate_of_closure/test_pyqt_flight_sample_inspector_rendered.py
           tests/rate_of_closure/test_pyqt_visualization_tab_visibility.py -q -n 0
 
       - name: Retain Playwright evidence

AGENT_HANDOFF.md

@@ -3,6 +3,32 @@
 > **Update this file with every PR and every push to main.**
 > Last updated: 2026-08-13
 
+## 2026-08-13 Flight hosted typing closure (#4433)
+
+Version 1.16.94 narrows the immutable NumPy return, keeps Qt signals and
+`eventFilter` on the concrete `FlightView`, and gives the execution mixin
+type-check-only optional result contracts. This closes the hosted Python 3.12 /
+MyPy 1.13 findings without runtime stubs or an MRO change. Flight authority,
+interaction, rendering, diagnostic evidence, and the open #4433 limitations
+are unchanged.
+
+## 2026-08-13 Flight synchronized sample inspector (#4433)
+
+Version 1.16.93 adds immutable, generation-bound accepted-flight bundles and
+matched React/PyQt raw-sample inspectors. The accepted exploration retains
+aligned time/position/velocity, while a maximum 1,002-sample plan binds exact
+current-primary time/position; the calm ghost
+is comparison-only. Twelve-CSS-pixel pointer picking and Arrow/Home/End/Escape
+navigation drive the sole 3D playback time without rerunning flight physics.
+React direct-entry and PyQt direct/delivery model, kernel, wind, launch,
+summary, origin, and landing evidence is validated before atomic publication;
+the separate target overlay is refreshed in the same PyQt transaction. Failed computation
+or UI/render publication retains prior scientific authority; if pixels cannot
+be restored the visible warning says the image may be stale. Three Chromium React
+selected viewport captures and eight PyQt selected/error window/canvas DPI PNGs are
+diagnostic-only, not approved goldens. Runtime-local sample indices are not a
+portable cross-solver identity, and #4433 remains open.
+
 ## 2026-08-13 Club Explorer hosted dtype closure (#4433)
 
 Version 1.16.92 explicitly types the structured binary-STL NumPy dtype and

SPEC.md

@@ -26,12 +26,42 @@
 | **Owner**               | D-sorganization                            |
 | **Primary Language(s)** | Python 3.11+, Rust, JavaScript, TypeScript |
 | **License**             | MIT                                        |
-| **Current Version**     | 1.16.92                                    |
-| **Spec Version**        | 1.16.92                                    |
+| **Current Version**     | 1.16.94                                    |
+| **Spec Version**        | 1.16.94                                    |
 | **Last Spec Update**    | 2026-08-13                                 |
 
 ## 2. Purpose & Mission
 
+### 2026-08-13 Flight hosted typing closure (#4433)
+
+Version 1.16.94 makes the accepted-flight NumPy snapshot return explicit,
+keeps Qt signals and event filtering on the concrete view, and supplies
+`TYPE_CHECKING`-only optional result fields to the execution mixin. Pinned
+Python 3.12 / MyPy 1.13 accepts all 13 changed Flight source files. Runtime
+science, atomic publication, interaction, artifacts, and open #4433 boundaries
+are unchanged.
+
+### 2026-08-13 Flight synchronized sample inspector (#4433)
+
+Version 1.16.93 adds matched React/PyQt inspection of exact runtime-local flight
+samples. React direct entry and PyQt direct/delivery each produce an immutable
+generation-bound bundle containing the complete producing context,
+launch/model/kernel/wind provenance, aligned time/position/velocity evidence,
+optional calm comparison, validated summaries, and a time/position plan capped
+at 1,002 samples before planner copying/allocation. The separate target overlay is
+transactionally refreshed from the accepted trajectory. Canonical tee origin,
+launch velocity, ground floor, landing,
+and wind-delta cohesion fail closed. The current primary cohort alone is
+selectable; the calm trace is a comparison ghost with no inferred raw-index
+correspondence. Twelve-pixel pointer selection and Arrow/Home/End/Escape update
+side/top markers, status, and the sole 3D playback timestamp without scientific
+recomputation. React and PyQt retain prior or honest empty authority on failure;
+PyQt publication spans target, renderer, controls, rows, deltas, statuses, and
+public references, while a failed pixel rollback is explicitly labeled stale.
+Three Chromium React selected viewport and eight PyQt selected/error window/canvas DPI PNGs are
+diagnostic-only, not approved goldens. Raw indices are runtime-local rather than
+portable solver identity, and broader #4433 tab/approval work remains open.
+
 ### 2026-08-13 Club Explorer hosted typing closure (#4433)
 
 Version 1.16.91 closes the pinned Python 3.12 / MyPy 1.13 Club Explorer

docs/development/RATE_OF_CLOSURE_CAMPAIGN_HANDOFF.md

@@ -1,5 +1,41 @@
 # Rate of Closure Campaign Handoff
 
+## 2026-08-13 #4433 Flight hosted typing closure
+
+- The accepted-flight NumPy snapshot has an explicit array return, Qt signals
+  and `eventFilter` remain on the concrete viewer, and optional accepted-result
+  fields exist only under `TYPE_CHECKING` in the execution mixin.
+- Pinned Python 3.12 / MyPy 1.13 accepts all 13 changed Flight sources; focused
+  Python/Qt behavior remains 36/36. This is static-only: science, interaction,
+  rendering, diagnostic evidence, and open boundaries are unchanged. SPEC is
+  1.16.94; #4433 remains open.
+
+## 2026-08-13 #4433 Flight synchronized sample inspector
+
+- A Python-owned golden and strict Python/TypeScript twins cap raw evidence at
+  1,002 samples before planner copying/allocation, preserve launch/apex/landing semantics,
+  and reject malformed, nonfinite, overflow-prone, translated, airborne-final,
+  below-ground, or provenance-incoherent trajectories.
+- React direct entry and PyQt direct/delivery each publish one immutable accepted
+  bundle with exact producing inputs, launch, model/kernel/wind provenance,
+  aligned time/position/velocity, optional calm comparison, derived summaries, and the
+  generation-bound time/position display plan. The separate target overlay is
+  transactionally refreshed from the accepted trajectory.
+  The calm ghost has no inferred correspondence and is never selectable.
+- Side/top pointer selection is bounded to 12 rendered pixels; keyboard
+  navigation and misses preserve focus and exact raw identity. Selection jumps
+  the sole 3D playback owner to the exact accepted timestamp without a solver
+  rerun. Canonical speed survives all presentation-unit switches.
+- PyQt commits target, view, controls, result rows, deltas, statuses, and public
+  references as one transaction. Failed repaint retains scientific authority;
+  failed pixel restoration is labeled stale/unavailable and idle redraw pauses
+  until a proven recovery. React retains prior or honest empty evidence with a
+  bounded alert and marks edited inputs as different from displayed context.
+- Three Chromium React selected viewport captures plus eight PyQt selected/error-prior
+  window/canvas DPI PNGs are diagnostic only. The retained raw grid differs by runtime,
+  so indices are runtime-local evidence, not cross-solver identity. SPEC is
+  1.16.93; #4433 remains open for remaining tabs and approval/AT coverage.
+
 ## 2026-08-13 #4433 Club Explorer hosted dtype closure
 
 - The structured binary-STL dtype is explicitly typed and three binding boolean

scripts/test_assertion_allowlist.txt

@@ -24,5 +24,7 @@ tests/rate_of_closure/pyqt_variation_render_probe.py
 tests/rate_of_closure/pyqt_variation_visual_state_probe.py
 # Subprocess entrypoint; semantic assertions live in the owning rendered test.
 tests/rate_of_closure/pyqt_putting_sample_inspector_probe.py
+# Subprocess entrypoint; semantic assertions live in the owning rendered test.
+tests/rate_of_closure/pyqt_flight_sample_inspector_probe.py
 # Subprocess entrypoint; semantic assertions live in the owning tab-visibility test.
 tests/rate_of_closure/pyqt_visualization_tab_probe.py

src/rate_of_closure/AGENT_HANDOFF.md

@@ -3,6 +3,32 @@
 > **Update this file with every PR and every push to main.**
 > Last updated: 2026-08-13
 
+## 2026-08-13 Flight hosted typing closure (#4433)
+
+- The immutable NumPy return is explicitly narrowed, Qt signal/event ownership
+  stays on the concrete `FlightView`, and the execution mixin's optional result
+  fields are declared under `TYPE_CHECKING` only.
+- Pinned Python 3.12 / MyPy 1.13 accepts all 13 changed Flight sources. This is
+  static-only: runtime MRO, scientific authority, interaction, rendering,
+  artifacts, and open #4433 limits are unchanged. SPEC is 1.16.94.
+
+## 2026-08-13 Flight synchronized sample inspector (#4433)
+
+- React direct entry and PyQt direct/delivery snapshot one complete accepted
+  flight: producing inputs, launch/model/kernel/wind provenance, aligned raw
+  time/position/velocity, optional calm comparison, summaries, and a bounded
+  time/position plan. The separate target overlay is transactionally aligned.
+- Only the current primary cohort is selectable. Exact raw-index selection is
+  generation-bound, uses a 12 CSS/display-pixel radius, retains on misses, and
+  synchronizes side/top markers, polite status, and the sole 3D playback time.
+- All target/view/row/delta/status/public-reference publication is transactional.
+  First failure stays empty; prior failure retains authority and selection. A
+  failed pixel rollback is reported as stale/unavailable rather than fabricated
+  retained imagery. Presentation-unit changes preserve canonical speed exactly.
+- Chromium React selected evidence covers 1440x900, 1280x720, and 390x844. Eight PyQt
+  selected/error-prior window/canvas PNGs cover DPI 1.0/1.5 with exact canvas identity. These
+  PNGs are diagnostic, not approved goldens. SPEC is 1.16.93; #4433 remains open.
+
 ## 2026-08-13 Club Explorer hosted dtype closure (#4433)
 
 - The binary-STL record now has an explicit structured NumPy dtype, and Qt/