TsProxy is not a CRD shipped by the Tailscale operator (v1.96.x).
The operator provides: ProxyClass, ProxyGroup, ProxyGroupPolicy,
DNSConfig, Tailnet, Connector, and Recorder.
Replace TsProxy with ProxyGroup (type: ingress) and annotate
rtx6000-brain-service with tailscale.com/proxy-group to expose
the vLLM endpoint via MagicDNS at rtx6000-brain-0.<tailnet>.ts.net:8000.
- Rename deployment to rtx6000-brain-vllm to match scaler target ref
- Set deployment replicas to 0 (KEDA controlled)
- Add pollingInterval: 10s and cooldownPeriod: 30s for faster response
This enables scale-to-zero when idle and quick scale-up on first request.
See https://github.com/sirius0xdev/gcloud-lab/tree/master/infrastructure%2Fgpus%2Fbase
