aula-15: implementação completa APM (Tempo + OTel + demo app)

Componentes:
- tempo-values.yaml: Grafana Tempo monolithic, 256Mi, 10Gi PVC
- otel-collector-values.yaml: recebe OTLP, exporta traces→Tempo,
  gera span metrics (RED)→Victoria Metrics via spanmetrics connector
- demo-app/: Node.js com rotas /fast (1 query), /slow (N+1, 51 queries),
  /fixed (JOIN), auto-instrumentado com OpenTelemetry
- alerts/latency-alerts.yaml: VMRule com Doherty threshold (p95>400ms)
- setup.sh: instala Tempo, OTel Collector, configura Grafana datasource,
  deploy demo app via ConfigMap (sem Docker build necessário)
- cleanup.sh: remove apenas recursos da aula-15, preserva aula-12

Zero hardcoded hostnames. Tudo via .env e placeholders.

aula-15/alerts/latency-alerts.yaml

@@ -0,0 +1,55 @@
+# VMRule - Alertas de Latência e Erros (APM)
+# Baseado em métricas geradas pelo spanmetrics connector do OpenTelemetry Collector
+# Métrica: http_server_request_duration_seconds_bucket
+
+apiVersion: operator.victoriametrics.com/v1beta1
+kind: VMRule
+metadata:
+  name: latency-alerts
+  namespace: monitoring
+  labels:
+    app: victoria-metrics
+spec:
+  groups:
+    - name: apm.latency
+      rules:
+        # Doherty Threshold: resposta acima de 400ms degrada a produtividade do usuário
+        - alert: DohertyThresholdExceeded
+          expr: |
+            histogram_quantile(0.95,
+              sum(rate(http_server_request_duration_seconds_bucket[5m])) by (le, service)
+            ) > 0.4
+          for: 5m
+          labels:
+            severity: warning
+          annotations:
+            summary: "P95 latency above Doherty threshold (400ms) for {{ $labels.service }}"
+            description: "Service {{ $labels.service }} has P95 latency of {{ $value | humanizeDuration }}. The Doherty threshold (400ms) states that productivity degrades when response time exceeds this limit."
+
+        # Taxa de erros HTTP 5xx acima de 1%
+        - alert: HighErrorRate
+          expr: |
+            (
+              sum(rate(http_server_request_duration_seconds_count{http_status_code=~"5.."}[5m])) by (service)
+              /
+              sum(rate(http_server_request_duration_seconds_count[5m])) by (service)
+            ) > 0.01
+          for: 5m
+          labels:
+            severity: warning
+          annotations:
+            summary: "Error rate above 1% for {{ $labels.service }}"
+            description: "Service {{ $labels.service }} has {{ $value | humanizePercentage }} error rate (5xx responses)."
+
+        # P99 acima de 1 segundo - latência crítica
+        - alert: HighRequestLatencyP99
+          expr: |
+            histogram_quantile(0.99,
+              sum(rate(http_server_request_duration_seconds_bucket[5m])) by (le, service)
+            ) > 1
+          for: 5m
+          labels:
+            severity: critical
+          annotations:
+            summary: "P99 latency above 1s for {{ $labels.service }}"
+            description: "Service {{ $labels.service }} has P99 latency of {{ $value | humanizeDuration }}. This indicates severe performance degradation."