Support per-profile priorities for GPU scheduling
https://kubernetes.io/docs/concepts/configuration/pod-priority-preemption/#priorityclass