Add more autoscaling options that can be defined per VM Scale Set

[marcoboffi]

Cluster autoscaler for Azure
Is your feature request designed to solve a problem? If so describe the problem this feature should solve.:

Currently it is only possible to define certain autoscaling options for the VM Scale Set using tags [link].
Having dedicated nodepools for specific workloads would be very useful to be able to differentiate as many parameters as possible for each nodepool.
In some cases I am forced to dispense with the autoscaler (scaling manually after alerts), while in other cases I have to execute the workload on different clusters.

Describe the solution you'd like.:

Add more autoscaling options that I can override to differentiate vmss autoscale behaviour for each nodepool.
In particular I'm interested to following parameters:

• new_pod_scale_up_delay
• max_unready_percentage
• max_unready_nodes
• scale_down_delay_after_add
• scale_down_delay_after_delete
• scale_down_delay_after_failure

Describe any alternative solutions you've considered.:

Additional context.: