perf(cmpp): add durable inbound fast path

tools/monitoring/cmpp-alerts.yml

@@ -182,7 +182,7 @@ groups:
   - name: cmpp-core-services
     rules:
       - alert: CmppCoreServiceInactive
-        expr: node_systemd_unit_state{name=~"cmpp-api\\.service|cmpp-gateway\\.service|postgresql\\.service|redis(-server)?\\.service|cmpp-minio\\.service|nginx\\.service",state="active"} == 0
+        expr: node_systemd_unit_state{name=~"cmpp-api\\.service|cmpp-send-worker\\.service|cmpp-gateway\\.service|postgresql\\.service|redis(-server)?\\.service|cmpp-minio\\.service|nginx\\.service",state="active"} == 0
         for: 2m
         labels:
           severity: critical
@@ -200,6 +200,16 @@ groups:
         for: 2m
         labels: { severity: critical, service: api }
         annotations: { summary: "API指标采集不可用", description: "Prometheus连续2分钟无法读取API内部指标端点。", currentValue: "{{ $value }}", threshold: "up = 1" }
+      - alert: CmppSendWorkerMetricsDown
+        expr: up{job="cmpp-send-worker"} == 0
+        for: 2m
+        labels: { severity: critical, service: send-worker }
+        annotations: { summary: "发送Worker指标采集不可用", description: "Prometheus连续2分钟无法读取独立发送Worker指标端点。", currentValue: "{{ $value }}", threshold: "up = 1" }
+      - alert: CmppInboundWorkflowBacklogCritical
+        expr: cmpp_worker_inbound_workflow_oldest_pending_age_seconds > 120
+        for: 2m
+        labels: { severity: critical, service: send-worker }
+        annotations: { summary: "CMPP耐久Inbox严重积压", description: "最旧待处理CMPP Inbox连续2分钟超过120秒。", currentValue: "{{ printf \"%.0f\" $value }}s", threshold: "120s" }
       - alert: CmppApiHttpErrorRateWarning
         expr: (sum(rate(cmpp_api_http_requests_total{status=~"5.."}[5m])) / clamp_min(sum(rate(cmpp_api_http_requests_total[5m])), 0.001) > 0.01) and (sum(rate(cmpp_api_http_requests_total{status=~"5.."}[5m])) / clamp_min(sum(rate(cmpp_api_http_requests_total[5m])), 0.001) <= 0.05) and sum(increase(cmpp_api_http_requests_total{status=~"5.."}[5m])) >= 5
         for: 5m