feat: improve scheduler recurrence, DB config, and docs

- Broaden scheduler query window to next N days for proper recurring event expansion (scheduler.py)
- Update DB connection logic for consistent .env loading and fallback (database.py)
- Harden timezone handling and logging in scheduler and DB utils
- Stop auto-deactivating recurring events before recurrence_end (API/events)
- Update documentation to reflect new scheduler, API, and logging behavior

.env.example

@@ -29,8 +29,11 @@ MQTT_KEEPALIVE=60
 # VITE_API_URL=https://your.api.example.com/api
 
 # Groups alive windows (seconds)
-HEARTBEAT_GRACE_PERIOD_DEV=15
-HEARTBEAT_GRACE_PERIOD_PROD=180
+# Clients send heartbeats every ~65s. Allow 2 missed heartbeats + safety margin
+# Dev: 65s * 2 + 50s margin = 180s
+# Prod: 65s * 2 + 40s margin = 170s
+HEARTBEAT_GRACE_PERIOD_DEV=180
+HEARTBEAT_GRACE_PERIOD_PROD=170
 
 # Scheduler
 # Optional: force periodic republish even without changes