In a k8s deployment the stored state is lost when a pod is
refreshed. When this happens the charm reports it is not ready
this is because the pools_available state has been lost. This
patch updates the stored state.
Tested here:
https://review.opendev.org/q/topic:test-ceph-client-interface
Change-Id: I5fff482e849b0d47ebc9b759a1559ccbad73af4d