Allow scaling down nodepools in 'out-of-capacity' situations

**Allow scaling down nodepools in 'out-of-capacity' situations during 'Update-AksHci'**:

**Description**:
If I have a cluster with too less memory left to create a new VM and start an `Update-AksHci` - the update will start but wait in an out-of-capacity loop until there is enough memory freed up to create a new VM.

One option is to scale down a nodepool, but scaling a nodepool is blocked in this situation:

Second option is to delete a nodepool, what succeeded and `Update-AksHci` continued.

**Feature Request**:
Please allow scaling down nodepools in out-of-capacity situations. We do not have anytime multiple nodepools available, where we can easily delete one, in case we run into an update hang because of running out of resources.