fix(reporter): fix and improve metric queries, improve data reliability
- Use last_over_time() with 10h window for all pgwatch_settings queries
- Fix version parsing to handle different PostgreSQL version string formats
- Add validation for empty setting names before processing
- Fix pg_stat_statements settings filter to use exact matching
- Add error logging when queries return no data
fix(grafana): update dashboard variable queries for reliability
- Replace pgwatch_settings_configured with pgwatch_db_size_size_b
- Use more reliable metrics for cluster, node, and database dropdowns
- Ensures dashboard variables populate correctly
fix(compose): simplify reporter invocation
- Remove manual cluster and node name extraction from instances.yml
- Reporter now auto-detects these from Prometheus metrics
- Reduces complexity and potential parsing errors
fix(docker): add version metadata to reporter image
- Add VERSION build arg requirement for proper versioning
- Add OCI image labels for better image metadata