[Feature] Concurrent node lifecycle operations during node pool updates

Today, while an AKS node pool is in a non-terminal state (Upgrading, UpgradingNodeImageVersion, Updating, or Scaling), customers cannot remove a specific bad node or add manual capacity without first cancelling the in-progress operation. This adds 20–60 minutes to incident response and pushes customers toward unsafe workarounds like direct VMSS manipulation.

With this capability, you can:

- Delete specific nodes with az aks nodepool delete-machines while the pool is upgrading, refreshing its node image, applying a configuration update, or being scaled by Cluster Autoscaler — on both manual and CAS-enabled pools.
- Manually scale up a non-CAS pool with az aks nodepool scale concurrently with an in-progress upgrade or update, via a new dedicated scale API that does not trigger full reconciliation. New nodes automatically use the target configuration, avoiding mixed-version pools.
- Track each operation independently through per-operation status (operation ID, percent complete, current step) and a new Active operations view in the portal and CLI.

Benefits:
- Meet node remediation SLAs without cancelling in-progress operations
- Respond to capacity spikes during upgrade windows without rollback risk
- Eliminate the need for unsafe direct-VMSS workarounds