Allow CSI drivers to detach volumes in case of down node (Tech Preview)

1. Proposed title of this feature request

Add non-graceful node shutdown to allow CSI drivers to detach volumes in case of down node

 2. What is the nature and description of the request?

When a node goes down non-gracefully, the CSI drivers do not have permissions to detach the disks from that node so it requires manual steps to detach those disks and let the pods using those disks be scheduled on a different node.

3. Why does the customer need this?

It causes downtime when a node goes down
It requires manual steps to be able to get the workload back to work again
It's a lack of high availability capabilities the product is expected to have

4. How would the customer like to achieve this?

The volumes should be detached from shutdown nodes, allowing their disks to be attached to a different node. STOR-744 scope is limited to the manual tainting approach. After a node is detected unheathy, operator has to taint the node.

5. For each functional requirement listed, specify how Red Hat and the customer can test to confirm the requirement is successfully implemented.

Ungraceful Shut down of a worker node with disks attached, taint the node, see if volume can be reattached, start the node again, remove the taint.

6. Is there already an existing RFE upstream or in Red Hat Bugzilla?

Yes: https://github.com/kubernetes/enhancements/pull/1116

https://github.com/kubernetes/kubernetes/issues/65392
https://github.com/kubernetes/kubernetes/pull/108486

 7. List any affected packages or components.

Red Hat OpenShift Container Platform.