fix: release Windows ACL temp-path memos

[Ingwannu]

Summary

• release successful and timed-out Windows ACL memos once an ephemeral file is proven absent
• key atomic-write timeouts by the stable destination so unique temp names cannot accumulate
• cover config/response state, OpenAI tier backups, management tokens, and tray replacements
• retain memo state when a hardened residual temp remains on disk

Part of #820. This intentionally extracts the confirmed Windows leak into a focused change instead of coupling it to #829.

Verification

• bun run typecheck
• bun run privacy:scan
• bun test tests/windows-secret-acl.test.ts tests/config.test.ts tests/server-management-auth.test.ts tests/windows-tray.test.ts (167 pass)
• full suite: 6427 pass / 6 skip; 5 unrelated GUI test import errors because this checkout lacks the GUI React dependencies (react, react/jsx-dev-runtime)

Compatibility

Stable destination/directory memo behavior and fail-closed ACL policy are unchanged. Only one-shot temp path entries are released after rename/unlink or proven ENOENT.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: ba90a6fd-d1a1-4dda-8456-3b3d7976d55f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}

[lidge-jun]

Maintainer triage: this is a valuable focused fix. It removes the confirmed Windows process-lifetime temp-path memo leak while preserving fail-closed handling when a residual file remains, and the focused coverage is broad across the affected write paths.

The branch currently conflicts with dev, so it needs a rebase. After that, an independent review from someone other than the author is required before landing. We are tracking this under #820 for the next cycle.

[lidge-jun]

Superseded by #892, which implements this defect's fix natively on top of the wave-1 retained-state framework (the PR's shape was used as a reference; see the PR description in #892 for the defect-specific deltas, including where the framework already covered part of it). Thank you for the contribution — the analysis in this PR drove the fix.