Description of problem:
When using the code in this PR to disable the Ingress capability in a HostedCluster, the CNO restarts periodically with the following error:
"if kind is a CRD, it should be installed before calling Start" err="no matches for kind \"IngressController\" in version \"operator.openshift.io/v1\""

How reproducible:
Always

Steps to reproduce:

1. Build a custom CPO image using the code from the PR above, and build a matching HyperShift operator image, and the Hypershift CLI binary.

1. Create a HostedCluster with Ingress disabled using:
    

   hypershift create cluster none \ --name=${HOSTED_CLUSTER_NAME} \ --base-domain=${BASEDOMAIN} \ --release-image=${OCP_RELEASE_IMAGE} \ --ssh-key=${HOME}/.ssh/id_rsa.pub \ --pull-secret=${PULL_SECRET_FILE} \ --disable-cluster-capabilities=Ingress \ --control-plane-operator-image=<CPO custom image>

1. Observe the CNO pod in the management cluster (hosted control plane namespace).

Actual results:
The CNO pod restarts repeatedly with the CRD-related error.

Expected results:
CNO should not attempt to watch or expect the IngressController CRD if the Ingress capability is disabled.

Additional info:
Related discussions:

1. https://redhat-external.slack.com/archives/C01C8502FMM/p1750879931748809
2. https://redhat-internal.slack.com/archives/CDCP2LA9L/p1750881335126359