Our test suite for joins is quite poor at the moment. Now that I look at it, it's generally just testing if it's hooked up and not if the exact results...

test_match_geopandas_series.py has proved to be super helpful for finding bugs in Sedona. We should replicate this for joins. In this recent PR fix to ST_Distance, there's some suspicion that it's possible distance join is buggy on an edge case. Comparing results with geopandas would be a great way to test this.

Afterwards, we should also improve our tests for the rest of sjoins and for sindexes.