fix(store): retry compaction when db not ready instead of waiting full interval

[corylanou]

Summary

Follow-up to #886. When a database returns ErrDBNotReady, the compaction monitor now retries after 1 second instead of waiting for the full compaction interval (which could be 24 hours for snapshots).

Changes

• Track when any DB returns ErrDBNotReady during compaction iteration
• Use short retry interval (1s) if retry needed, up to 30s timeout
• Log warning if timeout exceeded (DB never initialized)
• Reset retry counter and use normal interval after success

Behavior

Before: If DB not ready at startup, wait full interval (L9: 24h, L1: 30s, etc.)

After: Retry every 1 second until DB is ready (max 30s timeout), then use normal interval

Test plan

• [x] All existing tests pass (go test ./...)
• [x] Pre-commit hooks pass

Related to #877

🤖 Generated with Claude Code