Eliminates hanging on namespace deletion for OpenShift 4.3

[diverdane]

This change adds:

• A workaround for an OpenShift 4.3 bug (https://bugzilla.redhat.com/show_bug.cgi?id=1798282) whereby deleting a namespace may hang indefitely because sometimes finalizers are left undeleted on Kubernetes services in the namespace.
• A display of residual Kubernetes resources when namespace delete fails.

NOTE: Jenkins tests are currently failing on OpenShift 4.5. This failure is unrelated to the change being made here. The failures for OpenShift 4.5 appear could possibly be due to improper credentials being set for this routine in platform_login:

  elif [[ "$PLATFORM" = "openshift" ]]; then
    oc login $OPENSHIFT_URL \
      --username=$OPENSHIFT_USERNAME \
      --password=$OPENSHIFT_PASSWORD \
      --insecure-skip-tls-verify=true
    docker login \
      -u _ -p "$(oc whoami -t)" \
      $DOCKER_REGISTRY_PATH
  fi

Addresses Issue #125

[diverdane]

@hughsaunders agreed, removing finalizers is brute-force-ish, since we're removing ALL finalizers. Perhaps a little more well-behaved workaround would be to run the oc namespace delete --timeout ...., let it time out, and then remove residual finalizers. But, this would add 10 minutes to each run (based on the timeout value).

[diverdane]

approved, don't forget to change the branch in the git clone command

@orenbm thanks, I removed the clone of the temporary branch, since the cyberark/kubernetes-conjur-deploy change has been merged.