Scrape traefik gateway metrics into VictoriaMetrics

Both traefik instances already expose prometheus metrics on :9100 but
nothing scrapes them. Adds a VMPodScrape for traefik-system and enables
entrypoint/router labels for per-route dashboards.

apps/base/observability/kustomization.yaml

@@ -9,3 +9,4 @@ resources:
   - gateway.yaml
   - httproute.yaml
   - vmpodscrape-cnpg.yaml
+  - vmpodscrape-traefik.yaml

apps/base/observability/vmpodscrape-traefik.yaml

@@ -0,0 +1,23 @@
+---
+# Scrape metrics (:9100) from the traefik-internal and traefik-external gateway
+# pods. Picked up by the observability VMAgent (selectAllByDefault). The
+# app.kubernetes.io/instance pod label is kept as traefik_instance so series
+# from the two releases stay distinguishable.
+apiVersion: operator.victoriametrics.com/v1beta1
+kind: VMPodScrape
+metadata:
+  name: traefik
+  namespace: observability
+spec:
+  namespaceSelector:
+    matchNames:
+      - traefik-system
+  selector:
+    matchLabels:
+      app.kubernetes.io/name: traefik
+  podMetricsEndpoints:
+    - port: metrics
+      path: /metrics
+      relabelConfigs:
+        - sourceLabels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
+          targetLabel: traefik_instance

apps/overlays/au-syd1/traefik-system/values-external.yaml

@@ -29,6 +29,11 @@ providers:
     nativeLBByDefault: false
     labelSelector: "traefik.io/instance=external"
 
+metrics:
+  prometheus:
+    addEntryPointsLabels: true
+    addRoutersLabels: true
+
 logs:
   access:
     enabled: true

apps/overlays/au-syd1/traefik-system/values-internal.yaml

@@ -29,6 +29,11 @@ providers:
     nativeLBByDefault: false
     labelSelector: "traefik.io/instance=internal"
 
+metrics:
+  prometheus:
+    addEntryPointsLabels: true
+    addRoutersLabels: true
+
 logs:
   access:
     enabled: true