[BUG] Fix flaky test_spark_tuning

[miguelgfierro]

Description

The test tests/unit/examples/test_notebooks_pyspark.py::test_spark_tuning is not working properly. See: #1764 (comment)

According to @anargyri: One thing that is special with this test in particular is that we call ALS.fit() two or three times. So, I would speculate that this may "overload" the resources allocated by Spark in some way. I suggest you try separating the ALS training runs into separate tests and see if this fixes the issue. It's probably feasible to do this by parameterizing the notebook in a creative way.

In which platform does it happen?

How do we replicate the issue?

More details https://github.com/microsoft/recommenders/runs/7145344779?check_suite_focus=true

Expected behavior (i.e. solution)

Other Comments

[miguelgfierro]

duplicate of #1770