Container state monitoring (#2, @NotRetarded: "why keep this app strictly as just an updating tool?") — Docksentry now watches for state transitions and notifies on:
- a healthcheck turning unhealthy (and the recovery back)
- a container exiting with a non-zero code (zero exits stay silent)
- an OOM kill
- a crash + auto-restart (RestartCount bump — the case a plain exit check misses on
restart: alwaysfleets)
Anti-spam by design: transitions only, whole pass skipped while updates run (recreates never read as crashes), 30-minute flap cooldown, silent baseline on boot, quiet-hours/maintenance honored.
Config: MONITOR (default on), MONITOR_INTERVAL (60s). Per-container opt-out: docksentry.monitor=false. Fans out to Telegram/Discord/webhooks; 16 languages. Disk-threshold monitoring already existed (DISK_WARN_PERCENT).