Support cluster automatic ballance after an AZ is lost

Right now the operator will not restore the setup to be ballanced across multiple AZs in the case of an AZ failure. There has to be a manual intervention - either a rolling restart or by moving a leader node from an AZ with two leader nodes to teh AZ that came back from fail state. It would be nice if the operator was able to automatically rebalance the cluster.
(AZ = availability zone)