In order to switch from kube-core-operator renderer we need to make
sure that all manifests (bootstrap + phase 2) are provided from the
openshift kube/openshift operators.

All control plane operators from openshift implement a rendering
logic that lay down secrets, configs and YAML definitions for workload
resources as well as static pods for initial pivot.

In order to switch to OpenShift operators and bootstrap we
need to remove the kube-core-operator rendering and not
rely on the assets provided by this operator.

The new assets are provided by kube/openshift operator renderers.

Instead of hard-coding images in installer, use the origin-release
to extract the correct image names for bootstrap manifests.

Add the 'system:masters' into kubelet cert key so kubelet is able
to reach out to openshift kube apiserver successfully.

This is temporary and the "pod" team should figure out how to make
kubelet bootstrapping more secure using minimal permissions, preferably
using the 'system:node:*' service accounts to align with upstream.

In order to complete the bootkube start sequence we need to provide
checkpointer pod that has to be observed as "ready" by the bootkube.
Before these manifests were provided by kube-core operator and before
we have the checkpointer operator that provide these, we have to
install them directly temporarely.

The kube-proxy manifests are needed for the tectonic-network operator
in order to talk to the bootkube control plane. These should be owned
by networking team and moved to their operator.

bootkube: override control-plane configs – fix kubectl logs