Currently if multiple instances of the ceph-client chart are
deployed in the same Kubernetes cluster, the releases will
conflict because the clusterrole-checkdns ClusterRole is a global
resources and has a hard-coded name. This change scopes the
ClusterRole name by release name to address this.
Change-Id: I17d04720ca301f643f6fb9cf5a9b2eec965ef537
3b9aa44ac5