k8s/monitoring: add prometheus grafana loki observability

- Loki log aggregation (MinIO backed, 10-day retention)
- Promtail daemonset (pod + talos journal logs)
- Prometheus + kube-state-metrics
- Grafana dashboards (6-row template per service)

k8s/monitoring/dashboards/services/story-crater-backend.descriptor.yaml

@@ -0,0 +1,74 @@
+# story-crater-backend.descriptor.yaml
+# Build input for generating svc-story-crater-backend.yaml dashboard ConfigMap
+# and svc-story-crater-backend-rules.yaml PrometheusRule.
+# NOT applied to cluster — this is a reference for Claude's mechanical dashboard generation.
+
+service: story-crater-backend
+display_name: "Story Crater Backend"
+namespace: story-crater-backend
+service_type: message-worker
+folder: "Story Crater Backend"
+
+jobs:
+  - agent-worker
+  - check
+  - edit-collab
+  - canary
+
+metrics:
+  prefix: story_crater
+  up_selector: 'job=~"agent-worker|check|edit-collab|canary"'
+
+  rate:
+    metric: story_crater_messages_handled_total
+    labels: [agent, queue, status]  # status: succeeded | failed
+
+  duration:
+    metric: story_crater_message_handle_duration_ms
+    type: histogram
+    unit: ms
+
+  error:
+    metric: story_crater_app_metric_total
+    filter: 'severity="error"'
+
+  domain:
+    - metric: story_crater_queue_depth
+      kind: gauge
+      group_by: [queue]
+    - metric: story_crater_dedup_redeliveries_total
+      kind: counter
+      group_by: [agent, queue]
+    - metric: story_crater_outbox_publish_lag_ms
+      kind: histogram
+    - metric: story_crater_llm_tokens_used_total
+      kind: counter
+      group_by: [agent, kind]
+    - metric: story_crater_llm_call_duration_seconds
+      kind: histogram
+      group_by: [agent, model_provider]
+    - metric: story_crater_check_latency_ms
+      kind: histogram
+      slo_ms: 1200
+    - metric: story_crater_degrade_state
+      kind: gauge
+      group_by: [service]
+
+correlation_ids: [session_id, trace_id, tenant]
+
+logs:
+  loki_namespace_selector: 'namespace="story-crater-backend"'
+
+alerts:
+  - name: HighErrorRate
+    expr_template: rate_error_ratio
+    threshold: 0.05
+    for: 10m
+    severity: warning
+  - name: CheckLatencySLOBreach
+    expr_template: histogram_quantile_over_threshold
+    metric: story_crater_check_latency_ms
+    quantile: 0.95
+    threshold: 1200
+    for: 10m
+    severity: critical