Backport: Remove anyuid SCC requirement for OpenShift (#3813)

[curtbushko]

Manual backport of #3813 to release/1.2.x

• OpenShift does not want you to use a hardcoded user id and group when running containers.
• We currently hardcode 5995 and 5996 for the init and inject containers.
• OpenShift tells you what ids you can use by adding an annotation to the namespace you are deploying to.
• We now read that annotation from the namespace and change the ids at run time.
• This removes the requirement of having to run oc adm policy add-scc-to-group anyuid system:serviceaccounts:$TARGET_NAMESPACE when deploying to OpenShift

Changes proposed in this PR

• Added some helper code to connect-inject/common that reads from the namespace and gets the UID and Group to be used. This code has some fallbacks based from previous versions of OpenShift
• NOTE: The word 'range' is used in the OpenShift annotation but it is not really a range (anymore).
• If on OpenShift, use the IDs for init and connect inject.
• For CNI, we add an annotation to the pod with the redirect config. It includes the UID in the json blob. CNI uses this annotation/blob during Pod Network setup.
• Acceptance tests were creating the anyuid policy when running on OpenShift and that is not longer needed. (Files dropped/kustomize changes).

How I've tested this PR

• Unit tests
• Created a local CRC OpenShift cluster, deployed consul and static server to manually check the IDs.
• Had @wilkermichael run the connect acceptance tests against a real OpenShift cluster using my control plane image.

How I expect reviewers to test this PR

👀

Checklist

• Tests added
• CHANGELOG entry added