Currently, we're exercising external infra by using a single baremetal mgmt/infra cluster. We create a unique namespace for a guest cluster's VMs and then pass a the baremetal cluster's kubeconfig in as the "kubevirt-infra-kubeconfig".  This is cheating a bit because we're not actually running the HCP on a separate OCP cluster as the kubevirt VMs.

The result is we don't get an accurate signal of whether or not external infra works correctly. This setup can mask issues. For example, the bug CNV-29074 was only discovered once running external infra on a separate OCP cluster from the HCP. 

This task is to identify what options we have to accurately test external infra within openshift CI, and decide on an approach.