✨ Update Go module dependencies and enhance NATS consumer metrics. Add github.com/kylelemons/godebug as an indirect dependency. Upgrade github.com/prometheus/common to version 0.67.3. Change NATS consumer mode to "jetstream" in the development configuration. Improve observability by ensuring metrics for pending messages are initialized and recorded correctly in JetStream mode. Update documentation to reflect changes in metrics handling and consumer behavior.

docs/observability.md

@@ -32,7 +32,7 @@ The service exposes Prometheus metrics via the monitoring HTTP server (default `
   Count of general publish failures (not DLQ-specific), labelled by message category.
 
 - `caatsm_nats_consumer_pending_messages{stream,consumer}`  
-  Current pending message count for each JetStream consumer (useful for lag/backlog alerts).
+  Current pending message count for each JetStream consumer (useful for lag/backlog alerts). A zero sample is emitted only once at startup so the series exists; if JetStream stats queries fail later, the last known value is preserved rather than force-setting the gauge to `0`, which prevents false “queue cleared” alerts.
 
 Additional OTEL metrics are emitted via the configured OTEL endpoint, including:
 
@@ -282,4 +282,3 @@ Logging is done with Zap. The `internal/infra/log` package standardises fields v
 
 Handler and consumer logs should always be emitted through `WithMessageContext` to ensure these fields are present where applicable.
 
-