Race condition between volume reconstructor and volume discovery

[Andrzej-Zukowski]

Describe the bug
Race condition after node reboot. Kubelet triggers volume reconstruction procedure while csi-driver still performs discovery procedure.

Environment (please complete the following information):
Issue is Hardware and OS independent

To Reproduce

1. Shutdown the node
2. Update one or more DriverCRs to point to incorrect block storage. f.e. /dev/sda -> /dev/sdb. Alternatively change storage position in an enclosure
3. Boot the node
4. Wait till node is online
5. As soon as csi-baremetal-kernel-node pod is in creating stage isolate container outbound traffic to apiserver. Use whatever means suits you ( f.e NetworkPolicies ). Wait for updateDrivesCRs to fail. Immediately restore pods network connectivity
6. Wait till NodeStageVolume fails as it operates on either modified DriverCR or outdated DriverCR

Expected behavior
DriverCR shall reflect actual state of the world before volume reconstruction starts

Screenshots
Not applicable

Additional context
Not applicable