Needed so we can provide this workflow to customers following the proposal at https://github.com/openshift/enhancements/pull/1583

Reference https://issues.redhat.com/browse/WRKLDS-1015

kube-scheduler pods are created by code residing in controllers provided by the kubescheduler operator. So changes are required in that repo to add a toleration to the node-role.kubernetes.io/control-plane:NoExecute taint. 

https://github.com/openshift/cluster-kube-scheduler-operator/blob/4be4e433eec566df60d6d89f09a13b706e93f2a3/bindata/assets/kube-scheduler/pod.yaml#L13

The operator itself does not run in the control-plane nodes, but if that change is necessary it would be here: https://github.com/openshift/cluster-kube-scheduler-operator/blob/4be4e433eec566df60d6d89f09a13b706e93f2a3/manifests/0000_25_kube-scheduler-operator_06_deployment.yaml#L12