Autoscaling support

[jdheyburn]

Describe the solution you'd like

• When using PPS in Strict enforcementMode, when a pod in unschedulable from PPS perspective it will conflict with the Kubernetes cluster-autoscaler (CA)
• This can cause pods to be stuck in pending state, since CA will not scale up additional nodes
• CA uses the default-scheduler Filter logic to determine whether the pod can be scheduled on an existing node, not PPSs Filter

Offending logic in cluster autoscaler

Is there any support for autoscaling today, or perhaps in future? If not, is there perhaps a workaround instead?

Anything else you would like to add:
For our use case, we are looking for it to be Strict enforcement mode, since we want some control of how many nodes with a given label are deployed.

Environment: AWS, deployed via kops

• Kubernetes version: (use kubectl version): v1.21.14
• Placement Policy Scheduler plugins version: v0.1.0