[k8s] Run GPU labeller automatically on new nodes added to cluster

[romilbhardwaj]

From user:

Can whatever's happening in sky.utils.kubernetes.gpu_labeler be automatically applied when new nodes join the k8s cluster?
Capacity changes happen manually right now. It has just happened a few times when our team forgot to apply this step since most other k8s operators would automatically propagate to new nodes.

We could use daemonset to run on all nodes automatically as they are added, but that adds a small always-running CPU overhead for keeping the pod running on all nodes.

[github-actions]

This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[romilbhardwaj]

This mitigated by #3493.

[github-actions]

This issue is stale because it has been open 120 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[github-actions]

This issue was closed because it has been stalled for 10 days with no activity.