When CNO is managed by Hypershift multus-admission-controller does not have correct RollingUpdate parameterts meeting Hypershift requirements outligned here: https://github.com/openshift/hypershift/blob/646bcef53e4ecb9ec01a05408bb2da8ffd832a14/support/config/deployment.go#L81
There are two standard cases currently with hypershift: HA mode where there are 3 replicas spread across zones and then non ha with one replica. When only 3 zones are available you need to be able to set maxUnavailable in order to progress the rollout. However, you do not want to set that in the single replica case because it will result in downtime.
So when multus-admission-controller has more than one replica the RollingUpdate parameters should be