fix(viz): update admin port names (#15090)

**Problem**

After #14111, prometheus does not correctly scrape the control plane
pods due to a misconfigured regex. This is true for both the regular
control plane components as well as the multicluster addon.

**Solution**

Update the regex for the `linkerd-controller` and
`linkerd-multicluster-controller` jobs to keep any port that matches the
pattern `*.admin$` instead of matching the entire string `admin`. This
way, any port name that ends with or is just `admin` is matched. 

Additionally, any port that matches `linkerd-admin` is dropped to not
scrape the control plane proxies twice. This change is in line with the
handling in the `PodMonitors` for the control plane and multicluster
addon.

**Validation**

Set up two clusters with multicluster link enabled. Roll out the new
prometheus config and check that control plane metrics (not just the
control plane proxies) are present.

Fixes #14642 and #15089.

Signed-off-by: Benedikt Kittinger <benedikt.kittinger@post.at>

Author: bkittinger

viz/charts/linkerd-viz/templates/prometheus.yaml

@@ -68,7 +68,11 @@ data:
       - source_labels:
         - __meta_kubernetes_pod_container_port_name
         action: keep
-        regex: admin
+        regex: .*admin$
+      - source_labels:
+        - __meta_kubernetes_pod_container_port_name
+        action: drop
+        regex: linkerd-admin
       - source_labels: [__meta_kubernetes_pod_container_name]
         action: replace
         target_label: component
@@ -81,7 +85,11 @@ data:
         - __meta_kubernetes_pod_label_component
         - __meta_kubernetes_pod_container_port_name
         action: keep
-        regex: (linkerd-service-mirror|controller);admin$
+        regex: (linkerd-service-mirror|controller);.*-admin$
+      - source_labels:
+        - __meta_kubernetes_pod_container_port_name
+        action: drop
+        regex: linkerd-admin
       - source_labels: [__meta_kubernetes_pod_container_name]
         action: replace
         target_label: component