Fix Prometheus metrics handling for the operator.

The current state of the Helm chart is slightly confusing, because:
- There's no easy way to really disable prometheus metrics --
  `--enable-prometheus-metrics` defaults to true anthe current code
  either emits `--enable-prometheus-metrics=true` or nothing at all
  (making it `true` once again).
- The `http` container port is actually a `metrics` port (by convention
  from .e.g. webhook), but is present regardless of whether Prometheus
  metrics are enabled or not. To make it less confusing, this PR
  proposes renaming it.

Signed-off-by: Milan Plzik <milan.plzik@grafana.com>

keda/templates/manager/deployment.yaml

@@ -86,9 +86,9 @@ spec:
           - "--metrics-server-service-name={{ .Values.operator.name }}-metrics-apiserver"
           - "--webhooks-service-name={{ .Values.webhooks.name }}"
           - "--k8s-cluster-domain={{ .Values.clusterDomain }}"
+          - "--enable-prometheus-metrics={{ .Values.prometheus.operator.enabled }}"
           {{- if .Values.prometheus.operator.enabled }}
           - "--metrics-bind-address=:{{ .Values.prometheus.operator.port }}"
-          - "--enable-prometheus-metrics={{ .Values.prometheus.operator.enabled }}"
           {{- end }}
           {{- if .Values.opentelemetry.operator.enabled }}
           - "--enable-opentelemetry-metrics={{ .Values.opentelemetry.operator.enabled}}"
@@ -116,9 +116,11 @@ spec:
             failureThreshold: {{ .Values.operator.readinessProbe.failureThreshold }}
             successThreshold: {{ .Values.operator.readinessProbe.successThreshold }}
           ports:
-          - containerPort: 8080
-            name: http
+          {{- if .Values.prometheus.operator.enabled }}
+          - containerPort: {{ .Values.prometheus.operator.port }}
+            name: metrics
             protocol: TCP
+          {{- end }}
           env:
             - name: WATCH_NAMESPACE
               value: {{ .Values.watchNamespace | quote }}