Need to recreate the cluster during every test run

Some changes force us to recreate the cluster we're using in the workflow test.

For example the recent change from kube-system to kctf-system. Not recreating the cluster will mean we're having stale resources that can make it misbehave (e.g. a 2nd copy of the operator kube-system).

When running a test on one cl on branch A and then another on branch B, the two branches might have conflicting cluster changes and the 2nd test might either be a cluster upgrade or a cluster downgrade.

It sounds we would have to delete and recreate the cluster with an option to skip the step if the cluster has a version from which we know how to upgrade.

Similarly, there might be other objects in GCE left over from a different run (service account, load balancer) that could lead to issues. But I don't know if there's a good way to completely clean the project. But this might be less likely to make issues (?)

@sirdarckcat thoughts? so far it was okayish to handle this manually. But it can lead to issues in the future especially if two people develop at the same time.