Stop processing if pod fail to avoid blocking the CRD

[leleueri]

When a CRD is updated, the DC phase may be blocked in UPDATING is the Elassandra Pods (or Statefulset) can't start du to invalid configuration.

Due to the UPDATING state, a new Update of the CRD has no effect...

To avoid this situation we have to :

• keep a CRD copy after a Statefulset becoming RUNNING
• watch pod failure (or StatefulSet if the error comes from the STS)
• if the post fails to restart (define number of retry) during an UPdate phase:

• trigger a rollback of the STS to the previous CRD generation
• preserve the failed generationID in CRD status
• set the CRD Status in a specific status that requires a SysAdmin action to accept new CRD reconciliation