TestHAEtcdChaos runs as platform None.
This makes our input populating logic to not set the hostname in the service publishing strategy.
in AKS there's nothing to admit and reconcile those routes.
This results in no pods other than capi and cpo are running.
cpo rollout is blocked by

• lastTransitionTime: "2025-09-04T18:56:42Z"
  message: OAuth service route does not contain valid host; 8m since creation
  observedGeneration: 2
  reason: WaitingOnInfrastructureReady
  status: "False"

This issue was uncovered by this PR https://github.com/openshift/hypershift/pull/6738, which introduces a new validation for conditions.

Without that validation in place, the same scenario takes place but for some reason the test goes green anyways https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/test-platform-results/pr-logs/pull/openshi[…]e-clusters-gtlcj-ha-etcd-chaos-46gjq/core/pods/

The reason it goes green anyways I think is because their validations such as https://github.com/openshift/hypershift/blob/b68b3b8e98ad285d2996da8c09ea70e13b704c8e/test/e2e/chaos_test.go#L49 are just returning a func, but that's never executed.

Which means we are not running this validations at all anywhere, not even in aws.

We should change that test to actually run the validations and ideally run the HC as a real platform as well