fix(health/arrs): improve health check scheduling and arrs integration

[drondeseries]

Summary

This PR introduces improvements to the health check scheduling system and fixes a configuration requirement for Sonarr integration.

Changes

Health Check System

• Periodic Re-checks for Healthy Files: Modified GetUnhealthyFiles query to include files marked as healthy if their scheduled_check_at time has passed. This enables the intended tiered scheduling strategy (aggressive -> daily -> normal) to function effectively, ensuring long-term validation of file retention.
• Repair Retry Delay: Added a 1-hour backoff delay (scheduled_check_at = datetime('now', '+1 hour')) when updating status to repair_triggered. This prevents rapid retry loops when repairs fail.

Arrs Integration

• Optional Library Directory: Updated triggerSonarrRescanByPath to treat Health.LibraryDir as optional. Previously, missing this configuration would cause the repair process to error out. It now gracefully handles the missing config by falling back to other path resolution methods.

Code Cleanup

• Linter Fixes: Removed an unused sync.WaitGroup field in internal/importer/service.go.