Restore historical checks after the CI documentation update

[roger-sato]

Summary

The registered historical-snapshot check fails on current main because the tooling-ledger adapter rejects docs/ci-verification.md after merged #3139 appended its deadline instructions. Admit only that exact reviewed successor SHA-256, a3711fe12d560fac7726c530b9c8c622182f3d4f231a876c163616e49bd911bc. The old accepted hash remains allowed. Historical replay still authenticates and restores the original a854022e… document against the unchanged frontier741 manifest.

The added test uses the real adapter configuration, manifest and frozen document. It proves the temporary overlay contains the original bytes, leaves the current checkout unchanged, and rejects an unknown live append, a corrupted frozen document and a changed nested source pin. It runs through the existing registered package and watched CI test module.

Base: 20ecf2891d80b5acf18d980b2164fb44ea7faa90, priority revision 6. Tekalign's #3139 commit 54bbb81cc884819154776900830eeb4d9db4dc5e preserves the entire previous 8,297-byte document and appends 28 lines. This closes the document-admission issue explicitly left open by Arham Tahir's #3348 and our #3644; it preserves Dennis's historical adapter and all predecessor attribution. No frozen source, manifest, selected result or document is edited.

Test plan

Local environment: macOS arm64, CPython 3.12.14, Git 2.54.0.

• python3.12 -B scripts/research_package.py --package research/historical-snapshot-one-pass: before, 27 tests with 1 failure and 1 error at the document guard; after, all 28 tests pass. The new focused test also fails independently against the original adapter.
• python3.12 -O -W error -B -m unittest discover -s tests -p 'test_historical_check.py' -k ci_document_successor -v: passes, including all three rejection controls.
• python3.12 -B scripts/research_package.py --package research/isolated-test-bytecode and --package research/isolated-descendant-bytecode: every registered command passes; their declared outcomes remain completed / partial.
• Independent read-only review and independent focused regression pass. Frozen-document and coverage-manifest bytes match canonical main; git diff --check passes.

Follow-up verification on 2026-10-18: the fork tooling run passed this document guard, then failed on an existing omitted closure field. PR3686 addresses that separate checker failure. A complete isolated tooling-ledger invocation with this exact documentation admission and PR3686 passed the repaired boundaries, then stopped at the unchanged canonical PR3107 starting-head mismatch. PR3688 independently fixes the current-runner admission unit failure. Full ledger/CI success is not claimed; upstream Actions remain subject to maintainer approval. Fresh overlap review covered 354 open PRs, 2,964 closed PRs and 2,458 discussion records, plus targeted live searches; no equivalent repair was found.

Prepared for roger-sato with substantial OpenAI Codex assistance, including independent AI review. Apache-2.0.

Local integration follow-up (2026-10-19 JST): PR3692 repairs the initial-head mismatch using authenticated existing receipts. A disposable canonical-base tree combining its af040ab4 change with exact PR3648 a0d88d68, PR3686 932f3b1 (checker and tests), and PR3688 3463ba61 passes all 38 reconciliation tests normally and under -O, both with -W error, on macOS arm64 / CPython 3.12.14. One 600-second-capped tooling-ledger replay passed the repaired boundaries and batch1047, then stopped after 29.76 seconds at the separate canonical PR1048 detailed-review/evidence mismatch against historical batch1099. The untouched canonical checker independently reproduces that mismatch. Ledger, intake and frozen evidence remain unchanged; full ledger/CI success is still not claimed.