Implement the ability to expose separate metrics endpoints for each pod in HA control plane components. This ensures accurate counter metrics and pod-specific monitoring by eliminating the current limitation where metrics come from random pods in a service.

  Acceptance Criteria:
  - Each pod in HA control plane workloads has a separate metrics endpoint
  - Counter metrics maintain accuracy across multiple pods
  - Applies to kube-apiserver, kube-scheduler, kube-controller-manager, and other HA components
  - No breaking changes to existing metric collection