Data and Compatibility Invariants

This feature adds no entities, columns, indexes, or state machines. It changes Python paths and distribution metadata.

• Preserve tables, constraints, provider IDs, target identity hashing, stored scans, and notification eligibility behavior.
• Preserve every Alembic revision/down-revision/merge point. Update historical imports only as required; never squash or regenerate the revision graph.
• Resolve migrations from installed resources. Verify fresh SQLite and a synthetic earlier-revision database, alongside existing PostgreSQL checks.
• Preserve explicit Celery camply.* task names, JSON serialization, queues, arguments, and notification DTO defaults; Python paths become camply.worker.*.
• Preserve API schemas/authentication. Missing optional distributed dependencies use existing unavailable fields or HTTP 503, never false dispatch success.
• Preserve configuration names, SQLite data location, and database settings.

Record pre-existing SQLite migration or Python support failures separately. Do not redesign the database; implement the explicitly approved Python 3.11 support floor and 3.14 runtime defaults without changing application data.