Redesign sync job scheduling

- [ ] `cron`: Instead of running the first sync at startup, run at something like "every 20 minutes, starting at :00". Probably should also enable some jitter by default for the start time.
- [ ] ability to disable a job for a while after receiving an error
- [ ] persist retry state in db

The issue describes cron timing, temporary job disabling after errors, and retry-state persistence, but names no files, tests, or entry points. Start by locating the sync scheduler and its database state handling, then clarify the scheduling, jitter, disablement, and persistence behavior before implementation. Done means all three checklist items have defined and tested behavior.

Written by the indexing model from the issue text.