Fixed destination metrics warning when re-linking

Similar to #11246, the destination controller was complaning above
trying to register dupe metrics for the api cache gauges, when a given
target cluster got re-linked. This change unregisters the gauges for the
target cluster when said cluster is removed.

Supersedes #11252

Author: alpeb

controller/api/destination/watcher/cluster_store.go

@@ -173,6 +173,8 @@ func (cs *ClusterStore) removeCluster(clusterName string) {
 		return
 	}
 	r.watcher.removeHandlers()
+	r.watcher.k8sAPI.UnregisterGauges()
+	r.watcher.metadataAPI.UnregisterGauges()
 	close(r.stopCh)
 	delete(cs.store, clusterName)
 	cs.log.Infof("Removed cluster %s from ClusterStore", clusterName)

controller/k8s/metadata_api.go

@@ -100,6 +100,11 @@ func (api *MetadataAPI) Sync(stopCh <-chan struct{}) {
 	waitForCacheSync(api.syncChecks)
 }
 
+// UnregisterGauges unregisters all the prometheus cache gauges associated to this API
+func (api *MetadataAPI) UnregisterGauges() {
+	api.promGauges.unregister()
+}
+
 func (api *MetadataAPI) getLister(res APIResource) (cache.GenericLister, error) {
 	inf, ok := api.inf[res]
 	if !ok {