fix: stale Home lock recovery and Windows v1.0.1

[Dailin521]

Summary

• Refs support: 一直显示“操作执行中 另一项操作正在使用当前存储位置，完成后即可继续。” #95: distinguish verified stale Home locks from live or unverifiable owners. Read-only Status leaves lock files untouched; normal Apply revalidates and reclaims stale owners using the existing protocol.
• Verify every claim, canonical owner, file/directory identity and generation. Handle the Windows process-exit/start-time lookup race without treating permission errors as process exit.
• Keep unknown locks blocked with an actionable, distinct UI message; skip history-integrity scans when Diagnostics observes an active/unverifiable operation.
• Include the separately reviewed backup-list/prune read race fix and allowlisted Sync failure-stage/cause diagnostics.
• Prepare Windows Electron v1.0.1 (unsigned, manual installation). Generalize release guards only to strict 1.0.x patch releases; retain exact tag/SHA, full main CI, eight-asset and draft verification gates. No npm, Legacy, signing or automatic-install release.

Validation

• Local root tests: 531 passed, 54 platform/environment skips, zero failed.
• Architecture/workspace checks passed; Provider I/O gate: 47 passed, 14 platform skips.
• Additional lock/DTO regression checks: 31 passed, 2 file-symlink privilege skips on this Windows host.
• Production/full dependency audits: zero vulnerabilities.
• Root tarball smoke, Web/desktop production builds and release-version/notes checks passed.
• Windows packaged hidden-window E2E: all four scenarios passed, including stale lock → Preview → Sync → Restore, Provider switch/logging and UI layout (35.5 seconds).
• Initial CI caught a Node 16-incompatible test cleanup hook; corrected to the supported module hook and verified locally on Node 16.20.2. The original non-packaged UI run hit a screenshot-only timeout after its functional assertions; that site now uses the existing bounded hidden-viewport retry. No assertions were removed.
• Full current-head PR CI must pass before merge; final main CI and release-container evidence must pass before publication. Earlier-head CI is superseded, not reused.

No real user Codex Home was used. Tests use D-drive temporary fixtures. Reporter-specific root cause remains unconfirmed; keep #95 open for follow-up after release.

Architecture

ADR-0040/0041 and corresponding behavior/error/CLI contracts and fixtures updated. Provider first-line reads, eligible equal-byte in-place updates, timestamp restoration, backup-first/default two, native SQLite transactions and Restore recovery remain unchanged.

Release preflight follow-ups

• Addressed review comment 3986864987: both pre-build and pre-Draft checks read root/Desktop versions from the immutable tagged SHA. Matching and mismatching stable/RC inputs are covered; 15 release-guard tests passed.
• Fixed a C10 JSON Schema constraint that still accepted only 1.0.0 candidate versions. Eleven evidence tests passed, including actual-schema positive/negative patch candidate validation. The four-target index downloaded from CI run 34576428137 was also validated successfully with the corrected schema.
• That earlier run retained a Linux CDP timeout and Ubuntu dependency-download retry failure; its second attempt passed every business/platform test but exposed the schema constraint during final evidence generation. Those failures remain auditable. The latest commit reruns the complete PR gate; no failed or cancelled gate is treated as release evidence.
• After the same Linux CDP handshake-only timeout recurred, extended the existing Windows one-time clean-process activation retry to Linux. It applies only before browser connection, after verified cleanup, never to page or business assertions; both-platform rejection tests pass. Linux runtime dependency installation now matches the candidate job's Chromium-only dependency set. Final local Windows packaged run: 4/4 passed (2.5 minutes, including bounded hidden-viewport capture retries).

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-11T08:40:36.658151Z	7fdff10	New commits

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: cb06395859

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".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d6c8f4129e

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".