Complete bounded freeze audit and report unresolved per-path statistics

[GoGoKo699]

Scope

Repository-wide freeze audit based on 88d8e420ee292359cb5fb6c10869534c9e59cb03. This PR contains 20 changed files, four additions and no deletions. It introduces no new physical model, stronger universality claim, or release tag.

Findings closed

• Validate dimensions before coercion; reject booleans and fractional dimensions.
• Require finite entropy/energy logarithm bases greater than one.
• Compute finite entanglement-Hamiltonian gaps by subtracting logarithms rather than overflowing a ratio.
• Make normalized support/effective-rank coordinates zero at dimension one.
• Reject nonpositive/noninteger XXZ substeps and nonpositive/nonfinite record intervals instead of silently running an empty evolution loop.
• Mark numerically unresolved per-path variation before PCA or within-path Spearman correlation. A declared standard-deviation floor of 1e-10 excludes six nearly constant QCA_1 boundary paths from those statistics only. Eligibility and the floor are returned explicitly; global PCA and the chronology-control calculation are not altered.
• Synchronize scientific entry documents and the canonical claim registry with the chronology controls and negative results. Historical per-path estimates remain labeled v1.0.0 snapshots rather than current precision guarantees.

Verification already performed

Local expanded suite: 244 tests. Independent local probes: 1,935 assertions, including 585 high-precision entropy cases, 500 random spectra, analytical Schmidt states, partial-transpose identities, all 16 small-system dynamical conditions at n=4 and n=5, exact-exponential XXZ checks, and all 16,200 selected-spectrum pairs.

Pinned hosted production runs 34045124535 and 34046247153 passed the exact environment check, full controls (999 chronology draws and 199 matched-spectrum draws per law), full included-data analysis (3,000 bootstrap / 1,000 permutations), numerical-foundations verification, figure reproduction and the expanded test suite. The control scientific summary is unchanged; its archive is freshly regenerated with actual source provenance. A new pinned-hosted fresh-versus-historical statistics record is included.

The global boundary common mode remains 0.9026282301130131. The resolved per-path median is 0.9483959944036638 over 90 eligible paths. This is not a claimed improvement in physics; it avoids standardizing numerical noise. Fine tied-rank descriptors remain resolution-sensitive.

Preservation and limitations

Preserve original trajectory data, selected spectra, v1.0.0 result archives, public figures, dependency locks, licenses and existing tags. Full physical n=20 simulations and the complete large-system XXZ production were not rerun in this audit. Temporary staging workflows are absent from this PR and its ancestry. This is an AI-assisted numerical and consistency audit, not external journal peer review.

Require the normal PR checks and the separate post-merge main check to pass before treating the resulting commit as the freeze candidate.

[GoGoKo699]

Post-merge verification completed. Main is f2a2a69dd236efa1f6516b8664d55e5ce70d4e8c (tree 51e4b1ecf1455d2c47e721f0915b0529fc124a8f), and its own workflow run 34046870694 passed every stage. The normal PR run 34046738601 also passed. Dedicated pinned-hosted production runs 34045124535 and 34046247153 completed the full controls, included-data analysis and numerical-foundations verification before this merge. The expanded suite has 244 cases.

Independent local probes completed 1,935 assertions; two naive small-q boundary comparisons remain explicitly recorded and are explained only under their stated rounded-input allowance. A local variation-floor sensitivity check gives the same 90 eligible boundary paths and 270 eligible path-pair correlations at floors 1e-12, 1e-10 and 1e-8. This is a resolution sensitivity check, not a certified physical error bound.

The freeze recommendation is GO for this declared finite-data computational companion. Current secondary per-path statistics are distinguished from historical v1.0.0 snapshots. Full physical simulations through n=20 were not repeated in this audit; the conclusion is not an external peer-review certification. The v1.0.0 tag is unchanged at 90baf8c8124123a98ef1a654e14ca5ec83eae609. No new tag or release has been created.