[BEAM-5091] New wordcount integration test on DataflowRunner in Python 3

[markflyhigh]

This is an integration test that runs a simple version of wordcount pipeline and verifies output file content in Python 3.

Due to BEAM-6158, the test disables save_main_sessions flag. Since it duplicates from the existing wordcout example's behavior, we don't need to run it in existing Python 2 suite. Create Py3IT tag for integration tests that's available in Python 3.

Tests are done by running this test using DataflowRunner in Python 3.5.

@tvalentyn @aaltay @RobbeSneyders

---

Follow this checklist to help us incorporate your contribution quickly and easily:

• Format the pull request title like [BEAM-XXX] Fixes bug in ApproximateQuantiles, where you replace BEAM-XXX with the appropriate JIRA issue, if applicable. This will automatically link the pull request to the issue.
• If this contribution is large, please file an Apache Individual Contributor License Agreement.

It will help us expedite review of your Pull Request if you tag someone (e.g. @username) to look at it.

Post-Commit Tests Status (on master branch)

Lang	SDK	Apex	Dataflow	Flink	Gearpump	Samza	Spark
Go		---	---	---	---	---	---
Java
Python		---		---	---	---	---

[markflyhigh]

Run Python PreCommit

[markflyhigh]

Ready for review. @tvalentyn @aaltay

[tvalentyn]

Thanks, @markflyhigh. I support adding a wordcount test that does not require a main session for the purposes of improving existing Python 3 coverage. However I'd like us to have a better understanding whether we want to keep this test after BEAM-6158 is fixed. Possible reason to keep: in the future we may have another feature for which we will be adding test coverage gradually, and for which --save_main_session will not work out of the box. Reason not to keep - this test is a subset of regular wordcount test.

• If we want to keep this test, I suggest we modify https://github.com/apache/beam/blob/master/sdks/python/apache_beam/examples/wordcount_minimal.py to allow all pipeline options be passed as args, and write wordcount_minimal_it on top of that pipeline.
• If we don't want to keep this test, let's make it as easy as possible to switch current pipeline to a regular wordcount pipeline, while preserving the consistency of how tests are organized. We could think about it this way: how would we add a FnAPI Py3 integration test right now if we didn't have BEAM-6158. Would we put it in https://github.com/apache/beam/blob/master/sdks/python/apache_beam/examples/wordcount_it_test.py ?
• It looks like current test adds a new test folder, which I personally like, however now we will have two sets of tests in different places. I would suggest that we reorganize the suites in a different change, or move other tests to new folder in another commit on this PR.

[markflyhigh]

I would prefer to merge this test with test_wordcount_it once BEAM-6158 is fixed but for now, let's keep it in wordcount_it_test.py.

[markflyhigh]

@tvalentyn PTAL

[tvalentyn]

Thank you!