`auto-deploy delete` does not delete channel 2 postgres volume

When investigating some review app reliability issues with @devin, we found that review app deployments can fail on the DB_INITIALIZE or DB_MIGRATE step due to an older volume still existing in the namespace with conflicting changes.

Steps to reproduce

  1. Run auto-deploy delete
  2. Run kubectl -n $KUBE_NAMESPACE get pv or kubectl -n $KUBE_NAMESPACE get pvc

Expected result

There should be no output

Actual results

The resources are still present

Edited by Hordur Freyr Yngvason