[362] Pause cluster reconciliation during re-run

This MR brings changes to add the kube-unit to pause the cluster reconciliation during the re-run by setting the spec.paused field of the Cluster closes #362 (closed)

Finally there were few corner cases to address, as pause/resume jobs should not be enabled in few circumstances, and cluster healthchecks have to be adapted to handle the case where cluster is paused.

There also turned out to be some issue as kube-job is not properly designed to work in a namespace. It is deserving some cleanup, tracked in #422 (closed)