feat: detect permanently Pending pods (unschedulable, PVC issues)

[slauger]

Summary

Extend the operator to detect pods stuck in Pending state due to conditions that won't resolve on their own, such as:

• Unschedulable pods (insufficient resources, missing nodes, unmatched taints/tolerations)
• Failed PVC binding (missing StorageClass, no PV available)
• ProgressDeadlineExceeded on Deployments (new ReplicaSet can't come up

Motivation

These are common resource wasters alongside CrashLoopBackOff. A Pending pod with an unresolvable PVC claim can block storage resources indefinitely.

Notes

• This is a future consideration, not planned for the initial release
• Container failure states remain the primary focus
• Needs careful design to avoid false positives (e.g. pods that are temporarily pending during cluster scale-up)
  BODY
  )