fix: handle all 'how' values in df.join

[Thomzoy]

Fixes #248

[eakmanrq]

Thanks for contributing this @Thomzoy!

It would be great if we could at least get the left/right semi/anti cases added to the tests. Specifically the tests which demonstrate SQLFrame behaving the same as PySpark. See example here of a test showing that no on results in a cross join:

sqlframe/tests/integration/test_int_dataframe.py

Lines 882 to 896 in 250d86e

	def test_join_no_on(
	pyspark_employee: PySparkDataFrame,
	pyspark_store: PySparkDataFrame,
	get_df: t.Callable[[str], BaseDataFrame],
	compare_frames: t.Callable,
	):
	# No on results in a cross. Testing that "how" is ignored
	pyspark_employee = pyspark_employee.select(F.col("fname"), F.col("lname"))
	pyspark_store = pyspark_store.select(F.col("store_id"), F.col("num_sales"))
	employee = get_df("employee").select(SF.col("fname"), SF.col("lname"))
	store = get_df("store").select(SF.col("store_id"), SF.col("num_sales"))
	df = pyspark_employee.join(pyspark_store, how="inner")
	dfs = employee.join(store, how="inner")
	compare_frames(df, dfs)

Let me know if this is something you are willing to do and if not I will look into opening a new PR with the change + tests added.

[Thomzoy]

Included some test ! Another handled edge case: Spark treats a cross join with predicate as an inner join

[eakmanrq]

Included some test ! Another handled edge case: Spark treats a cross join with predicate as an inner join

Nice! The different implicit behavior of Spark is hard to match so thanks for fixing that!

[eakmanrq]

@Thomzoy Looks like there is a merge conflict. Can you fix that and we will get this in?