fix(windows): reap republished daemons during nuke

[nsxdavid]

Summary

• Follow-up to merged Add Windows warm daemon IPC #641 for the Windows nuke cleanup path.
• Reap metadata-discovered Windows warm daemons in bounded passes so retry-mode serve/mcp daemons that republish cli-daemon.pipe after the visible owner exits are also terminated.
• Keep cleanup scoped to cli-daemon.pipe records under {home}/.codedb/projects and retain process-image verification before termination.

Issue

No standalone issue exists; this addresses the post-merge Codex review note about multiple retry-mode Windows daemons.

Validation

• zig build --global-cache-dir .zig-global-cache test-mcp -Dtest-filter="nuke" --summary all - 7/7 tests passed
• zig build --global-cache-dir .zig-global-cache test --summary all - 869/873 tests passed (4 skipped)
• zig build --global-cache-dir .zig-global-cache --summary all - 3/3 steps succeeded
• PYTHONUTF8=1; python scripts/e2e_mcp_test.py --binary zig-out/bin/codedb.exe --project . - 20/20 passed on Windows using a repo-local test home
• git diff --check - passed

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 989f8eaf45

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".