Virtual Router is not properly cleaned up when workload cluster deployment fails

[aleixrm]

Reproduced using tilt. If you follow the tilt development guide, when setting the tilt-settings.yaml file in the cluster api repository, just forget to enable the bootstrap and control plane providers:

default_registry: "localhost:5000"

provider_repos:
- /local/path/to/repo/cluster-api-provider-opennebula

enable_providers:
- opennebula-infrastructure

extra_args:
  core:
    - "--v=4"
  opennebula-infrastructure:
    - "--v=4"

debug:
 opennebula-infrastructure:
   continue: true
   port: 30000

Then execute make tilt-up in the cluster-api repository, and deploy a kubeadm workload cluster executing make one-apply in the CAPONE repo. It will fail. Then execute make one-delete in the CAPONE repo for destroying the cluster and you will see that the control plane vRouter has not been deleted.