Prometheus: watch CRs in all namespaces

This allows application to create ServiceMonitors in their own namespace
.
Fits much better with our least privilege and multi-tenancy approaches.

On top of that remove the need to add a label to each ServiceMonitor.
This makes them easier to use.

Note that
any: true
instead of
matchLabels: {}
as described here:
prometheus-community/helm-charts#106
fails, because not in schema.

Author: schnatterer

applications/cluster-resources/monitoring/prometheus-stack-helm-values.ftl.yaml

@@ -162,6 +162,22 @@ grafana:
 
 prometheus:
   prometheusSpec:
+    # Find podMonitors, serviceMonitor, etc. in all namespaces
+    serviceMonitorNamespaceSelector:
+      matchLabels: {}
+    # With this, we don't need the label "release: kube-prometheus-stack" on the service monitor
+    serviceMonitorSelectorNilUsesHelmValues: false
+    podMonitorNamespaceSelector:
+      matchLabels: {}
+    podMonitorSelectorNilUsesHelmValues: false
+    ruleNamespaceSelector:
+      matchLabels: {}
+    ruleSelectorNilUsesHelmValues: false
+    scrapeConfigSelectorNilUsesHelmValues: false
+    probeNamespaceSelector:
+      matchLabels: {}
+    probeSelectorNilUsesHelmValues: false
+      
 <#if podResources == true>
     resources:
       limits: