Fix Google Calendar archive cleanup reliability

[martin-forge]

Fixes #1695.

What this changes

• await Google Calendar cleanup during toggleArchive() instead of firing it and forgetting it
• make deleteTaskFromCalendar() return success/failure and only clear googleCalendarEventId after a confirmed delete or a 404/410
• preserve the pre-archive event ID across archive moves so deletion still has the identifier it needs
• keep auto-archive queue items pending when the task is archived locally but Google cleanup still has to be retried
• retry Google cleanup for already-archived tasks that still have a lingering Google Calendar link

Why

The previous flow could archive a task locally while the Google event remained behind, and on some failure paths the local googleCalendarEventId could be cleared anyway. That leaves an orphaned event with no local pointer for retry.

This PR makes the archive flow retry-safe: local archive can still succeed, but the event ID is preserved until Google deletion actually succeeds, and the auto-archive queue keeps retrying cleanup while the link remains.

Tests

• npx jest tests/unit/services/TaskService.test.ts --runInBand
• npx jest tests/unit/issues/issue-google-calendar-archive-reliability.test.ts --runInBand

I also validated the patch on a clean upstream copy after generating the release notes bundle required for typecheck/build in this repo.