ENH - Update node pool usage for ClearML

[viniciusdc]

This was first spotted by @leej3 in #984, where he identified that some of ClearML inner resources didn't have a specific node selector key assigned to them, leading to some of those having weird behaviors when scaling for node usage, e.g. they could be deployed at wherever node available in the referred namespace disregarding the correct resource tags (in this case ClearML).

• There was a PR to fix the mentioned issue, but due to the work made on Initial commit for auth and stages workflow #1003, we weren't able to apply those changes.

• I am opening this issue to create a report of that PR and track any new changes made after v0.4.0+ to address this issue as part of the Remove ClearML, Prefect and kbatch as core services #1217 work.

[aktech]

We dropped support for ClearML