Force MQTT pushall re-sync after a report gap (fixes #14)

The printer publishes partial deltas most of the time; the accumulator
merges each one onto whatever state it already has. When the printer
powers off, our cloud MQTT connection stays up (it's a persistent
session to Bambu's broker, not a socket to the printer), so nothing
signals the outage. When the printer reconnects and resumes its normal
partial updates, those deltas get merged onto the stale pre-outage
state — fields the delta doesn't mention (e.g. nozzle_temp) stay frozen
at their old values. Sending any command happens to trigger a full
report from the printer, which is why toggling the light "fixes" it.

Track the time of the last message per BambuPrinter instance. When a
new message arrives after a gap longer than
BAMBU_RUN_MQTT_RESYNC_GAP_SECONDS (default 90s), request a pushall
before processing it, forcing a complete state refresh instead of
trusting the partial delta to overwrite stale fields.

bambu_run/conf.py

@@ -81,5 +81,12 @@ class _Settings:
     def CLOUD_SYNC_DAYS(self):
         return get_setting("BAMBU_RUN_CLOUD_SYNC_DAYS", 30)
 
+    # Seconds of silence on the MQTT report topic that, once broken by a new
+    # message, is treated as "the printer was probably offline" and triggers
+    # a pushall re-sync instead of trusting the partial delta to fill in stale
+    # fields left over from before the gap.
+    @property
+    def MQTT_RESYNC_GAP_SECONDS(self):
+        return get_setting("BAMBU_RUN_MQTT_RESYNC_GAP_SECONDS", 90)
 
 app_settings = _Settings()