merging docker health monitor into main

[shark1Martin]

No description provided.

[wfr-data-acquisition]

Summary of changes in this PR

Original work (shark1Martin)

• Added health-monitor Docker sidecar that uses the Docker socket to inspect container states and writes metrics to InfluxDB monitoring database (monitor.container, monitor.service, monitor.ping measurements)
• Exposed /api/health-status endpoint on the data-downloader API so the DAQ website can poll it
• Updated allowed CORS origins

Review fixes (wfr-data-acquisition)

• Removed dual-write from /api/health-status: the endpoint was writing to InfluxDB on every frontend poll (every 15s). The health-monitor sidecar is the authoritative writer at its own interval — no need to write from the API too
• Added depends_on: data-downloader-api to health-monitor in docker-compose so it doesn't start before the API is ready
• Replaced runs_per_minute metric with last_scan_duration_seconds — measures how long the slicks scan actually took, which is meaningful for monitoring
• DAQ website updated to display scan duration: "OK · last scan Xs" on the Telemetry Scanner card
• Added missing env vars to .env.example for health-monitor config

Data Downloader improvements (wfr-data-acquisition)

• Multi-season table fix: each season now uses its own table name (WFR25 → WFR25, WFR26 → WFR26) instead of a single global INFLUX_TABLE. The old code would query the wrong table for past seasons
• Past season manual scans: UI now shows a season dropdown next to "Trigger Scan" so you can force a scan of any past season (e.g. WFR25)
• Scheduled scans scope: periodic worker (6AM daily) only scans the active season (highest year). Past seasons are manual-only
• Fixed PR bug in services.py: removed dead code block that referenced non-existent self.settings.influx_database attribute — would have caused a runtime crash on the second mark_finish call
• Ghost run cleanup: merge_scanned_runs was keeping ALL vanished entries forever, causing ~150 artifact entries from the old pre-slicks scanner (non-round-hour timestamps, 1-row windows) to persist indefinitely. Now only preserves vanished entries that have a user note
• Bumped slicks to >=0.2.1
• Fixed CI typo: workflow had slicks>=2.0.1 instead of slicks>=0.2.1

[haoruizhou]

Congrats on your first shipped feature @shark1Martin

[haoruizhou]

good to go