CSI NodeStage stuck on connecting after a Failover

[tiagolobocastro]

Describe the bug
When a volume has gone through failover and the target sits on another node, the uri in volumeattachment is incorrect.
In this case, if volume is NodeStaged then it will keep failing...

To Reproduce
Steps to reproduce the behavior:

1. create deployment pinned to an io-engine labelled node with a 2-replica volume
2. remove io-engine label from target node (target is moved to another node)
3. reset the node where deployment is pinned to
4. when node is back, pod will not go to running state

Expected behavior
Pod should go to running state

Additional context
As the CSI publish_context is immutable, this means we need to sync up with the openapi during a node stage...