feat: add cpu and memory to gpu e2e quota

[skel84]

Summary

• extend the GitOps-managed gpu-e2e ClusterQueue to cover cpu and memory as well as example.com/gpu
• update the managed smoke CronJob to request cpu and memory explicitly
• keep the same test-flavor so Kueue accounts for all three resources in the same staging queue

Validation

• kubectl apply --dry-run=server -k kubernetes/apps/gpu-e2e/gpu-e2e
• confirmed direct apply was immediately reverted by Argo because gpu-e2e is GitOps-managed on main
• this PR is required to validate the cpu/memory queue contract under real GitOps state