Implement CCCMO render to run CCM on bootstrap node

This allows master nodes to be initialized with CCM replica from
bootstrap, speeding up process of cluster installation.

Without CCM running on bootstrap node, most of the cluster operator
and operand pods would have to tolerate new taint and a different
configuration of node.kubernetes.io/not-ready (NoSchedule) taint:

- effect: NoSchedule
  key: node.cloudprovider.kubernetes.io/uninitialized
  value: "true"
- effect: NoSchedule
  key: node.kubernetes.io/not-ready

Without this bootstrap addition or taint toleration the cluster with CCM
would never become ready.

To test this feature on AWS build installer and deploy on AWS.

To create a cluster run:
```bash
./hack/build.sh
./bin/openshift-install create manifests
cat <<EOF > manifests/manifest_feature_gate.yaml
---
apiVersion: config.openshift.io/v1
kind: FeatureGate
metadata:
  annotations:
    include.release.openshift.io/self-managed-high-availability: "true"
    include.release.openshift.io/single-node-developer: "true"
    release.openshift.io/create-only: "true"
  name: cluster
spec:
  customNoUpgrade:
    enabled:
    - ExternalCloudProvider
    - CSIMigrationAWS
    - CSIMigrationOpenStack
  featureSet: CustomNoUpgrade
EOF

OPENSHIFT_INSTALL_RELEASE_IMAGE_OVERRIDE=quay.io/dgrigore/release:cccmo-render ./bin/openshift-install create cluster
```

data/data/bootstrap/files/usr/local/bin/bootkube.sh.template

@@ -51,6 +51,7 @@ CLUSTER_ETCD_OPERATOR_IMAGE=$(image_for cluster-etcd-operator)
 CONFIG_OPERATOR_IMAGE=$(image_for cluster-config-operator)
 KUBE_APISERVER_OPERATOR_IMAGE=$(image_for cluster-kube-apiserver-operator)
 KUBE_CONTROLLER_MANAGER_OPERATOR_IMAGE=$(image_for cluster-kube-controller-manager-operator)
+CLUSTER_CLOUD_CONTROLLER_MANAGER_OPERATOR_IMAGE=$(image_for cluster-cloud-controller-manager-operator)
 KUBE_SCHEDULER_OPERATOR_IMAGE=$(image_for cluster-kube-scheduler-operator)
 INGRESS_OPERATOR_IMAGE=$(image_for cluster-ingress-operator)
 
@@ -230,6 +231,36 @@ then
     record_service_stage_success
 fi
 
+if [ ! -f cloud-controller-manager-bootstrap.done ]
+then
+    record_service_stage_start "cluster-cloud-controller-manager-bootstrap"
+	echo "Rendering Cloud Controller Manager core manifests..."
+
+	rm --recursive --force cloud-controller-manager-bootstrap
+
+	# Copy the CCCMO images configMap to resolve pod images from internal registry in CCCMO render run
+	# Source config map is located in https://github.com/openshift/cluster-cloud-controller-manager-operator/blob/master/manifests/0000_26_cloud-controller-manager-operator_01_images.configmap.yaml
+	bootkube_podman_run \
+	--name copy-cloud-controller-manager-images \
+	--rm -it \
+	--entrypoint /bin/bash "${RELEASE_IMAGE_DIGEST}" \
+	-c "cat /release-manifests/0000_26_cloud-controller-manager-operator_01_images.configmap.yaml" > manifests/ccm-images.yaml
+
+	bootkube_podman_run \
+		--name cloud-controller-render \
+		--volume "$PWD:/assets:z" \
+		"${CLUSTER_CLOUD_CONTROLLER_MANAGER_OPERATOR_IMAGE}" \
+		/render run \
+		--images-file=/assets/manifests/ccm-images.yaml \
+		--dest-dir=/assets/cloud-controller-manager-bootstrap \
+		--cluster-infrastructure-file=/assets/manifests/cluster-infrastructure-02-config.yml
+
+	cp cloud-controller-manager-bootstrap/bootstrap/* bootstrap-manifests//
+
+	touch cloud-controller-manager-bootstrap.done
+    record_service_stage_success
+fi
+
 if [ ! -f kube-scheduler-bootstrap.done ]
 then
     record_service_stage_start "kube-scheduler-bootstrap"