[Phase 10] State Recovery & RDS Migration

[kination]

Description

Enhance orchestrator fault-tolerance by rebuilding state and reconciling with remote worker resources.

Tasks

• Ensure all Task.Process state transitions write to DB audit log
• Implement startup recovery: rebuild ETS from DB on restart
• Reconcile pending/running tasks against Worker Plane on restart (re-subscribe K8s Watch, poll Lambda)
• Add idempotency key to task dispatch to prevent duplicate execution
• Migrate from SQLite to RDS PostgreSQL (Multi-AZ)
• Update Ecto repo config for RDS