Add a `--skip-teardown-upon-failures` flag allowing deep dive in the environment to uncover the root cause of an issue

[lavocatt]

Potential usages for the Github CI of the operator:

- name: Dump Diagnostics
  if: failure()
  run: |
    mkdir -p diagnostics
    kubectl get events -A > diagnostics/events.txt
    kubectl logs -n my-namespace -l app=my-operator > diagnostics/operator.log
    kubectl get pods -A -o wide > diagnostics/pods.txt
    # ... specific describes ...
- name: Upload Artifacts
  if: failure()
  uses: actions/upload-artifact@v4
  with:
    name: diagnostics-dump
    path: diagnostics/