This is a followup for CNV-29003 in which we've enabled the NodePoolUpgrade test, but just for replace strategy, in which a node pool release image update triggers a machine rollout resulting in creation of new nodes with the updated RHCOS/kubelet version.

With InPlace strategy, the machines are not getting recreated, but are getting upgraded only with soft-reboot. We've noticed it doesn't work for kubevirt as expected, as the updated node is getting stuck at SchedulingDisabled and the nodepool.status.version is not updated.

This task is for find the root cause, fix and make the InPlace nodepool test pass on the presubmit CI.