Add VictoriaMetrics observability stack + sync catalog for monitoring test

- VM stack 10 charts: victoria-metrics-cluster/auth, victoria-logs-cluster,
  victoria-metrics-agent/alert, opentelemetry-collector, kube-state-metrics,
  prometheus-node-exporter, alertmanager, perses (JWT/OIDC, Infisical-ready)
- ArgoCD ApplicationSet (syncWave) + per-chart dip-values overlays
- doc/victoria-metrics-architecture.md, define-chart-resources updates
- includes pending working-tree changes (mlflow, kubeflow, apisix, CLAUDE.md)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

manifests/helm/victoria-logs-cluster/0.1.5/custom-values.yaml

@@ -0,0 +1,45 @@
+# =============================================================================
+# VictoriaLogs Cluster — PaaSup 커스텀 오버라이드
+# 차트: victoria-metrics/victoria-logs-cluster
+# 멀티테넌시: AccountID 기반 스토리지 격리 (단일 노드와 달리 실제 격리)
+#   - 쓰기: vlinsert:9481  /insert/{AccountID}/opentelemetry/v1/logs
+#   - 조회: vlselect:9471  /select/{AccountID}/logsql/query
+# 로그 수집은 opentelemetry-collector가 VictoriaLogs-AccountID 헤더로 전송한다.
+# =============================================================================
+
+vlinsert:
+  replicaCount: 1
+  resources:
+    requests:
+      cpu: 100m
+      memory: 128Mi
+    limits:
+      cpu: 500m
+      memory: 256Mi
+
+vlstorage:
+  replicaCount: 1
+  retentionPeriod: 7d           # 로그 보존 기간. 운영 규모에 맞게 조정
+  persistentVolume:
+    enabled: true
+    size: 5Gi
+    # storageClass: "longhorn"
+  extraArgs:
+    memory.allowedPercent: "60"
+  resources:
+    requests:
+      cpu: 100m
+      memory: 256Mi
+    limits:
+      cpu: 500m
+      memory: 512Mi
+
+vlselect:
+  replicaCount: 1
+  resources:
+    requests:
+      cpu: 100m
+      memory: 128Mi
+    limits:
+      cpu: 500m
+      memory: 256Mi