Add resilient reminder retry monitoring

[szyunlin]

Summary

Implements resilient reminder background-job retry and monitoring for #130.

• Adds persisted retry state to reminders: retry count, max retries, next retry time, last attempt, last error, and failed/dead-letter state.
• Updates /reminders/run so one provider failure no longer blocks other due reminders. Successful reminders are marked sent, transient failures are scheduled with bounded 5/15/45 minute backoff, and exhausted jobs move to a failed state.
• Adds /reminders/jobs for retry/dead-letter monitoring and POST /reminders/{id}/retry for manual reset.
• Updates PostgreSQL schema, OpenAPI docs, README notes, and backend coverage.

Verification

• sh ./scripts/test-backend.sh tests/test_reminders.py passed: 4 tests.
• sh ./scripts/test-backend.sh passed: 24 tests.
• docker compose run --rm backend sh -lc "black --check app tests && flake8 app tests" passed.

/claim #130

[szyunlin]

Verification update for #130:

• Focused reminder tests passed locally: sh ./scripts/test-backend.sh tests/test_reminders.py -> 4 passed.
• Full backend suite passed locally: sh ./scripts/test-backend.sh -> 24 passed.
• Formatting/lint passed locally: black --check app tests && flake8 app tests.

The implementation keeps the existing reminder flow, adds persisted retry/dead-letter state, exposes /reminders/jobs monitoring, and provides POST /reminders/{id}/retry for manual recovery.

[szyunlin]

Short demo recording for #130: https://asciinema.org/a/z9lb8UXrBvML7I77

It shows the committed change summary and the focused reminder retry tests passing (4 passed).