When the control plane is upgraded, the CVO is upgraded and it reports completion according with some operators ability to run their operands in the data plane, e.g. message: Cluster operators console, dns, image-registry, ingress, node-tuning, storage are not available. Fine.

Now HCP wants to emit a signal to communicate a version rollout of the components running management side has completed [1]. HCP can infer that easily for components it manages directly. But not for operands which run management side and are managed by those components, e.g. ovn control plane, snapshot controller.

The goal of this epic is to create a contract to avoid this [2] workaround. Contract:

• Any operand running on a container/Deployment management side must have a label "hypershift.openshift.io/managed-by: <operator-name>".
• Any operand running on a container/Deployment management side must signal their current version with annotation, e.g.: "release.openshift.io/version: 4.19.0".
• A version is current for a Deployment when it has completed a rollout or at least one container is running on the image version and the rest are running on >= version.

[1] https://github.com/openshift/hypershift/pull/6300
[2] https://github.com/openshift/hypershift/pull/6492