Proposed title of this feature request

We should start work to replace prometheus-adapter with metric-server

What is the nature and description of the request?

Metrics Server is a scalable, efficient source of container resource metrics for Kubernetes built-in autoscaling pipelines. Currently prometheus-adapter serves this role in our stack. This has several drawbacks that a switch to metric-server would allieviate.

• lower monitoring stack complexity
• improve resource usage of prometheus, since metric-server has no need to query prometheus directly

Why does the customer need this? (List the business requirements)

This will indirectly affect customers by improving performance and resiliency of our stack

List any affected packages or components.

kube-prometheus, CMO

This will initially need upstream work before we can benefit from this in CMO.