CI cleanup task deletes CRD before we then immediately create it again
Summary
In our CI run, we remove the CRD as part of our cleanup script, but we also run that cleanup before deploy if the previous deploy was a failure. This results in us removing our CRD before install, and causes errors.
This is currently blocking a clean pipeline run of master: https://gitlab.com/gitlab-org/charts/gitlab/-/jobs/401529420