Move timelock alert delivery to the continuous worker #439

Epic: #437

Outcome

timelock notifications arrive continuously through one production sender.

Scope

• Reuse existing decoding, formatting, protocol routing, and internal Yearn mirroring.
• Preserve complete scheduleBatch operation grouping; do not alert on incomplete result pages.
• Queue processing and notification work so AI explanations and Telegram cannot block ingestion.
• Persist delivery state and retry failures without repeating successfully delivered messages
  during normal reconnects. Document the remaining ambiguous-send behavior of the Telegram API.
• Cut over after pilot validation, retiring the hourly sender task and its obsolete cursor
  handling. Keep HTTP reconciliation in the worker and update monitoring metadata/runbooks.

Acceptance criteria

ordinary and batch operations match current alert content/routing, delivery failures
and restarts recover correctly, and only the continuous worker sends timelock alerts.

Dependencies

Blocked by #438. Validate coverage and latency before changing production alert delivery.

Related work

• Current implementation: https://github.com/yearn/monitoring/blob/main/protocols/timelock/timelock_alerts.py
• Current schedule: https://github.com/yearn/monitoring/blob/main/automation/jobs.yaml

Include relevant behavior/failure tests and operational documentation in the implementation. Update monitoring metadata when monitor behavior changes.