ETCD Learner Mode went to Beta in k8s 1.29 and is now default enabled.

When a new stacked etcd instance comes up, it joins the cluster in learner mode.

The API Server cannot perform rpc calls against its etcd instance, and fails to come online.

In theory the etcd instance should be promoted to a full member and this issue should
be resolved, but for some reason this is not happening.

While we investigate a root cause, we are disabling this new feature flag explicitly.