Fix flaky pyspark.sql.dataframe.DataFrame.writeStream doctest by stopping streaming query

https://issues.apache.org/jira/browse/SPARK-47199 didn't fix the flakiness in the pyspark.sql.dataframe.DataFrame.writeStream doctest : the problem is not that we are colliding on the test but, rather, that the test is starting a background thread to write to a directory then deleting that directory from the main test thread, something which is inherently race prone.

The fix is simple: stop the streaming query in the doctest itself, similar to other streaming doctest examples.