The test image can use credentials and kubeconfig for both manager and managed (or to-be-managed) cluster, with the same fallback as Jenkins: use to-be-managed kubeconfig when the managed cluster is not yet bound so setup can perform fleet management if needed.
 
As a CI/CD maintainer, I want the test Dockerfile to support mounting or providing two kubeconfigs (manager + managed/to-be-managed) and the same fallback behavior as Jenkins, so that containerized Fleet Virtualization test runs work in dual-cluster environments and setup can bind the cluster when needed.