Support cleanup of machines even when the target APIServer is not reachable

[prashanth26]

How to categorize this issue?

/area ops-productivity /kind epic /priority 3

What would you like to be added: MCM to be able to support the cleanup of machines even when the target (shoot) APIServer is not reachable.

Why is this needed:

• This would help in cases where we would like to force delete machines deployed for a cluster where bringing up the APIServer seems to be causing issues.
  • We can have scenario during hibernation/deletion of shoot cluster, the API server is gone before machine obj could be deleted, and we lead to the same issue. In such cases the VM deletion is not attempted as of today as we have 2 pain points
    • we are not able to do drain (even when force delete is enabled for machine obj, as that also tries to force delete pods and talk to the target API server)
    • we are not able to delete node obj

But we would not want this to be automated, but ops triggered (by putting annotation) , so that we don't clean the backing VM without customer consent.

[prashanth26]

cc: @vlerenc @dguendisch

[AxiomSamarth]

/assign @AxiomSamarth

[gardener-robot]

@himanshu-kun Label priority/normal does not exist.