fix: address late-landing bot review findings across 11 merged PRs

[joeharris76]

Pull Request

Description

Consolidated fix for chatgpt-codex-connector review comments that landed on 11 already-merged PRs after they merged (late-landing findings): #1032, #1037, #1035, #1030, #1031, #1028, #1025, #1024, #1022, #1020, #1019. Each source PR's review thread will be replied to and resolved citing this PR's commit.

• qpc-05/fix(qpc-05): surface plan capture/load failures instead of converting them to absence #1032 (P2): thread the real DataFrame plan-capture-error cause through to both the .plans.json companion (existing_errors=) and the main result row (plan_capture_error field); reset per-run capture-error state on ExpressionFamilyAdapter so it doesn't leak across runs on a reused adapter instance.
• chore(todo): file 5 query-plan-capture review follow-ups (qpc-13..17) #1037 (3 findings): fix qpc-13's scope_limit path (_spark_helpers.py lives directly under benchbox/platforms/, not benchbox/platforms/dataframe/); add the missing deps.needs: [qpc-05] to qpc-13 and qpc-15 planning YAMLs.
• docs(todo): add a standing human/maintainer action queue #1035 (P1): gate human-action-required.yaml's branch-rename admin action (w2) on the default-branch switch (w3) too, matching the source TODO's A-before-B sequencing.
• fix(release): release-cut curation survives untracked paths; refresh uv.lock #1030 (P1): release-cut's git commit now passes --no-verify, since the curation step already deletes .pre-commit-config.yaml and _project/scripts/ (the local hook entrypoints) from the working tree before the commit runs.
• chore(todo): add DuckLake platform TODOs (core adapter + catalog/storage backends) #1031 (P2): fix python -m benchbox platform list → uv run -- benchbox platforms list (plural, actual console script) across both DuckLake TODO planning files.
• fix(qpc-03): fingerprint v2 — encode tree shape, escape fields, honest integrity + versioning #1028 (P1+P2): correct plan_changed against the full tree-walk fallback result so a pure fingerprint-encoding-version bump (v1→v2) is never misreported as a changed plan; thread fingerprint_version through PlanMetadata/PlanHistory so a version-boundary crossing is excluded from flapping/version-history diffs instead of always reading as "changed".
• fix(qpc-08): make plan history real — attribute fixes, add_run wiring, platform-partitioned flapping #1025 (P2): normalize a "Z"-suffixed timestamp to "+00:00" before datetime.fromisoformat(), which only accepts "Z" from Python 3.11 onward — the project's floor is 3.10.
• fix(qpc-07): to_dict-first plan serialization, anonymize plans companion, exact-basename path arithmetic #1024 (P1): recursively strip physical_operator.platform_metadata across the full operator tree for anonymized exports, not just the top-level raw_explain_output — many parsers copy raw EXPLAIN text (paths/hosts/usernames) into per-node metadata under platform-specific key names (details, extra_info).
• fix(qpc-02): stop corrupting status/iteration/run_type on load, fix capture_time 0.0 drop #1022 (P2 × 2): match each repeated query failure to its own errors[] entry (FIFO per query_id) instead of always reusing the first; synthesize a standalone FAILED row for any error not actually consumed by a queries[] row, not just for query_ids entirely absent from queries[].
• feat(ci): scheduled orphaned-commit detector (stranding w1) #1020 (P1+P2): document orphaned-commit-detector.yml as a third instance of the known "scheduled workflow absent from default branch" class in repo-admin-settings.md (the workflow already follows the safe ref: develop-checkout shell pattern, so no code change is needed there — only the admin default-branch-switch action, already tracked); use the latest PR (by number) as the branch-diff cutoff in detect_orphaned_commits.py and skip the branch when that latest PR closed without merging, so an abandoned closed PR's own commits aren't misreported as orphans.
• chore(todo): reconcile qpc-01 (landed) and correct qpc-03 (still open) tracker state #1019 (P2): move the qpc-01 planning YAML from TODO/ to DONE/ (already marked status: Completed).

Type of Change

• Bug fix
• Documentation
• New feature
• Breaking change
• Refactor / chore

Testing

Every fix has a dedicated regression test, verified to fail before the fix and pass after via git stash:

uv run -- python -m pytest tests/unit/platforms/dataframe/ tests/unit/core/results/ tests/unit/core/query_plans/ \
  tests/unit/core/manifest/ tests/unit/test_release_infrastructure.py tests/unit/test_results_exporter.py \
  tests/unit/workflows/ -q
# 2262 passed, 69 skipped

Full tests/unit/ -m fast suite (23,274 tests) also run: 3 pre-existing failures, all confirmed to reproduce identically on a clean develop checkout (unrelated to this PR — sandbox running as root bypasses two filesystem-permission-error tests, and the zstd CLI binary is absent for a third).

uv run --project _project/scripts -- python _project/scripts/todo_cli.py check-graph passes (no dangling references / cycles) after the TODO YAML edits and the TODO→DONE move.

ruff check / ruff format --check / ty check all clean on every changed file.

Public Contract Check

• I updated docs/reference/public-contracts.md, or this PR does not change public, beta-public, deprecated, generated, experimental, or repo-only contract surfaces.
• I checked result schema fields, MCP tool parameters, platform support status, wrapper facades, adapter subclassing hooks, and generated docs for contract-map impact. PlanMetadata.plan_fingerprint_versions and QueryResultInput.plan_capture_error are additive fields; existing .plans.json/main-file consumers are unaffected.
• Any platform/benchmark count claim in docs is generated/checked from registry metadata, or explicitly marked editorial/non-authoritative.

Documentation

• I updated the relevant user-facing docs (docs/operations/repo-admin-settings.md).
• I added regression notes for behavior changes (inline comments explain each fix's "why").
• I confirmed API contract wording remains accurate.

Code Quality

• I ran formatting and lint/type checks.
• I reviewed impacted modules for backwards compatibility and migration impact.
• I added/updated tests for behavior changes and error paths.
• I confirmed public contracts and release checklists are still valid.

Artifact Hygiene

• I did not commit raw screenshots, browser reports, generated logs, benchmark outputs, or temporary binary artifacts.
• No binary/raw evidence files added.

Notes

Not CODEOWNERS-gated (no changes under benchbox/core/query_plans/parsers/**, benchbox/core/expected_results/**, or the other soundness-critical paths) — squash auto-merge will be enabled.

🤖 Generated with Claude Code

---

Generated by Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: de98320f9c

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".