Is this a bug report or feature request?

• Feature Request

What should the feature do:
Currently, when managing multiple CephClusters in the same cluster, Rook will reconcile them serially. If there is some issue with one of the clusters (e.g. the osds never complete reconcile), reconcile of other clusters can be blocked. As a workaround until the health of the failed cluster can be fixed, Rook should allow skipping of the failed cluster to continue reconcile of the other clusters.

This is related to #13848 where we are looking at how to better implement the operator with multiple instances, but at least this small change will provide a workaround in the meantime.

We already implement the label ceph.rook.io/do-not-reconcile on ceph daemons, we can easily also implement this label on the cephcluster CR and just skip the reconcile at the top level.

What is use case behind this feature:
With multiple clusters managed in the same cluster, allow reconcile of clusters even if one cluster is blocked.