updated kubeflow installation script so that the helm command is idempotent

[prli]

Which issue is resolved by this Pull Request:

running make deploy-kubeflow sometimes fails due to various reasons. After fixing whatever issue that cause the failure, I am unable to run the script again.

Error: INSTALLATION FAILED: cannot re-use a name that is still in use
Traceback (most recent call last):
  File "utils/kubeflow_installation.py", line 256, in <module>
    install_kubeflow(
  File "utils/kubeflow_installation.py", line 79, in install_kubeflow
    build_component(
  File "utils/kubeflow_installation.py", line 132, in build_component
    install_helm(component_name, installation_path, namespace)
  File "/Volumes/workplace/kubeflow/kubeflow-manifests/tests/e2e/utils/utils.py", line 222, in install_helm
    assert install_retcode == 0
AssertionError
make: *** [deploy-kubeflow] Error 1

Description of your changes:

Updated the helm command from helm install to helm upgrade --install, which would run an install if a release by this name doesn't already exist.

https://helm.sh/docs/helm/helm_upgrade/

Testing:

• [N.A] Unit tests pass
• [N.A] e2e tests pass
• Details about new tests (If this PR adds a new feature)
• Details about any manual tests performed

Manually tested by running deploy-kubeflow. Observed that installation is successful without Error: INSTALLATION FAILED: cannot re-use a name that is still in use.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[akartsky]

Thank you for your contribution

[surajkota]

Thanks @prli