IIUC, before using hard affinities for HA components, we needed this to avoid scheduling problems during upgrade.

See https://github.com/openshift/cluster-monitoring-operator/pull/1431#issuecomment-960845938

Now that 4.8 is no longer supported, we can get rid of this logic to simplify the code.

• I don't know if no longer supported also applies to 4.8 that may want to upgrade someday. (in this case, we'll never be able to get rid of the code.)

• Maybe keep this mechanism to influence upgrades somewhere (maybe git history is sufficient), we may need to re-use it in the future.

This will reduce technical debt and improve CMO learning curve.