Provide failback method to manage group replication

[paulomach]

Bug Description

In case juju fails to operate the cluster, there's no way to manage the cluster due roles being revoked.

To Reproduce

n/a

Environment

n/a

Relevant log output

n/a

Additional context

No response

[paulomach]

In this case, there's still the cluster_admin user to fallback to, so it can be workarounded.

[taurus-forever]

@shayancanonical is it still topical?

[shayancanonical]

I am not sure what juju failing to operater the cluster mean. If there's an error in initial set up after the privileges for the root user are downscaled, the cluster will be new (without data) so we could use juju to delete the application and try again. I'm not sure what the motivation for this issue is (is it a race condition that we ran into?)

[taurus-forever]

@paulomach please share your issues better please, if topical we need to move it to Jira to pickup the next sprints and fix. Tnx!