Fix Kubernetes resource creation retries

[atanasdinov]

The previous retry mechanism would false positively return successfully when a single manifest file contains multiple resources and falls under the following conditions:

• at least one fails with "AlreadyExists" error
• all remaining ones fail with a different error

The new approach will instead parse the output line by line to verify that such occurrences are not possible.

[ldevulder]

Do you thing that this PR could be merged this week or next one? As this is a blocker for the multi-nodes openQA test.

[dbw7]

Tested across a variety of VM configurations with our single node deployment example that deploys multiple helm charts and uses the local-path-provisioner manifest that has multiple resources. Everything seemed to deploy properly across each of the VM configurations. I believe we should be fine to rebase and merge once it's marked as ready for review. I can take over and do that if needed

[atanasdinov]

@dbw7 Thanks for verifying; there are no conflicts so unless there has been a specific change related to Kubernetes spin-up recently, we should be good to approve and merge as is.