[Feature Request]: Rewrite pytest markers to avoid including ML test suites into a catch-all postcommit suite.

[tvalentyn]

What would you like to happen?

Current test configuration causes confusion for authors of new suites and makes some test suites run twice (see discussions on: #28571 #28534)

Issue Priority

Priority: 2

Issue Components

• Component: Python SDK
• Component: Java SDK
• Component: Go SDK
• Component: Typescript SDK
• Component: IO connector
• Component: Beam examples
• Component: Beam playground
• Component: Beam katas
• Component: Website
• Component: Spark Runner
• Component: Flink Runner
• Component: Samza Runner
• Component: Twister2 Runner
• Component: Hazelcast Jet Runner
• Component: Google Cloud Dataflow Runner

[AnandInguva]

if a test is using it_postcommit along with a ML framework marker, it is considered to be an IT test. These IT tests then will be possibly run on Dataflow with the respective requirements file.

We provide both the markers because we want to only collect IT tests of that ML framework but not unit tests.

[damccorm]

I don't think this should be a release blocker. The problem in question is that tests are running more often than they should, which doesn't represent risk to users.

@tvalentyn I'm removing the release blocker, if you disagree feel free to add it back