Upgrades dependency to use a more compatible version of spark

The new spark integration relies heavily on the new pandas UDF API (3+).
3.2+ also brings in pandas by default, so I htink it makes sense to
update the dependency. If someone wants, they can always bypass it or
reach out to us.

setup.py

@@ -74,7 +74,7 @@ def load_requirements():
         "dask-diagnostics": ["dask[diagnostics]"],
         "dask-distributed": ["dask[distributed]"],
         "ray": ["ray>=2.0.0", "pyarrow"],
-        "pyspark": ["pyspark[pandas_on_spark]", "pandas<2.0"],  # I'm sure they'll add support soon,
+        "pyspark": ["pyspark>3.2", "pandas<2.0"],  # I'm sure they'll add support soon,
         # but for now its not compatible
         "pandera": ["pandera"],
     },