Audit scheduled and batch jobs

Audit these scheduled jobs.

Per job:
1. **If it fails, does anyone find out?** Not "is it logged" — is there an alert, and does it reach a person. *Silent failure is the normal state of scheduled work and the reason this audit exists.*
2. **Is it idempotent?** What happens if it runs twice — from a retry, a manual re-run, or overlapping schedules.
3. **What happens if a run is still going when the next starts?** Does it queue, overlap, or corrupt state.
4. What happens if it is skipped entirely — does the next run catch up, or is that window lost forever.
5. Runtime versus interval. A job averaging close to its interval is already failing intermittently.
6. Timezone and daylight-saving behavior for anything on a daily or nightly schedule.
7. Dependencies between jobs that are enforced only by the clock, which is not enforcement.

Then:
- Jobs with no failure detection. Fix these first.
- Jobs whose schedules collide, especially around midnight and month-end.
- Jobs nobody can explain the purpose of. *Propose stopping them and watching what breaks*, which is faster than tracing them.
- What should not be a cron job at all — event-driven work on a timer.