Add check/warning for pre-existing ClusterRole in Kubernetes example scripts

[diverdane]

Is your feature request related to a problem? Please describe.

If you are using the authn-k8s demo example in examples/kubernetes interatively,
and you do the following sequence:

• Run the kubernetes authn-k8s demo
• Delete the demo namespace (e.g. kubectl delete namespace app-test)
• Instead of deleting the Helm release (e.g. with helm delete conjur-oss),
  you simply delete the Conjur OSS namespace (e.g. with kubectl delete namespace conjur-oss)
  (This isn't the documented way to clean up, but hey, it happens.)
• Re-run the demo (e.g. with ./start)

Then you will see errors during Helm upgrade saying that you already have a
ClusterRole object.

Describe the solution you would like

In the examples/kubernetes/2_helm_install_or_upgrade_conjur.sh script, add
a check for a pre-existing ClusterRole, and if there is one, perhaps
display a warning and set the chart value to not create the cluster RBAC.

Describe alternatives you have considered

Additional context