* Working with Spark 3.5.0

* Use SparkSession instead of deprecated SQLContext on the python side

* Correction in init

* Update tests for new spark session configuration

* Set spark log level to error

* Run all tests with and without codegen

* Re-name

* Update tests to include product of all the things that were originally tested

* Tests with no backward rows to match with

* Add tests for empty dataframes

* Add tests for preserving nulls in the input

* Add tests for preserving nulls when tolerance is used

* Fix codegen__left_join_some_rows_have_no_backwards_match

* Fix interpreted__left_join_right_dataframe_is_empty

* Fix interpretted__inner_join_searching_backward_for_matches_with_tolerance

* Tidy

* Remove debug `.show()`s

* Start tests

* Valid tests

* Fix codegen version

* Tests with duplicate rows

* Delete unnecessary code. It was a legacy of the normal join's multi matching behaviour

* Working inner joins with more code removed and tolerance applied inside main search loop

* Working inner join

* Tidy

* Very minor clean ups

* Remove some unnecessary branching

* Fix schema assertion for left_join_duplicate_join_keys

* Remove unneeded bound condition

* Fix schema assertions for nulls in join keys

* Update comments and rename variables

* Auto-format

* Remove slightly misleading comment

* Remove completed TODO comment

* More comment adjustments

* Remove `condition` from exec code

* Fail physical planning if there are non-equi conditions

* Add test

* Throw an exception

* Initial testcase

* Correct test

* Attempt to create a new version of normal `.join`

* Delete mess

* Switch to new spark extensions framework

* First explicit pitJoin compiles

* In progress: create new entrypoint to PIT join

* Working test

* All scala tests working

* Create python side joinPIT to replace context

* Rename some paramaters

* Update parameter order in tests

* Orgnise imports

* Tidy

* Make spark session an explicit argument

* Update Readme