Relax dependency on specific 'k8gb' namespace name

[ytsarev]

• We are relying on operator namespace to be named 'k8gb' in the helm chart templates and in the codebase
• It is not really flexible and goes against https://github.com/operator-framework/community-operators/blob/master/docs/best-practices.md
• To fix that we might want to use the approach similar to https://github.com/helm/charts/pull/22378/files

[ytsarev]

Just found another motivation to fix this one.
If one does helm install in another namespace, e.g. default helm --kube-context kind-eu install k8gb k8gb/k8gb --set k8gb.clusterGeoTag=eu a bad thing(tm) happens

$ k get pod
NAME                                                        READY   STATUS    RESTARTS   AGE
etcd-cluster-h7nqx8gslr                                     1/1     Running   0          4m19s
etcd-cluster-hhdv5x2sxv                                     1/1     Running   0          3m30s
etcd-cluster-wzhrpxx9vz                                     1/1     Running   0          2m49s
k8gb-coredns-657c46d8c5-l7vzp                               1/1     Running   0          5m16s
k8gb-etcd-operator-etcd-backup-operator-6fdbbff98f-6292d    1/1     Running   0          5m16s
k8gb-etcd-operator-etcd-operator-7b58798b-n2qwj             1/1     Running   0          5m16s
k8gb-etcd-operator-etcd-restore-operator-5cc6b5b46c-dgx82   1/1     Running   0          5m16s

$ k -n k8gb get pod
NAME                            READY   STATUS    RESTARTS   AGE
external-dns-5cbb5c6697-zh vwnd   1/1     Running   0          5m32s
k8gb-5d66f649bd-8gspr           1/1     Running   0          5m32s

Pod allocation is getting totally wrong between hardcoded k8gb and a current default depending on k8gb subsystem