Using cap-ci by importing your own GKE clusters is broken

[viccuad]

Hello,

We realized versions of the cap-ci pipeline that consume already created GKE clusters from the pool, instead of creating its own, are broken. This happens because we depend on GKE_CLUSTER_NAME, etc. The fix is SUSE/catapult#242.

Posting here since we need to make sure in the pipeline that all that is needed for importing clusters is a kubeconfig (and general public cloud creds that are already there as secrets).