Add an example for having two columns as the grouping key in group aggregate pandas UDF

https://github.com/apache/spark/pull/20295 added an alternative interface for group aggregate pandas UDFs. It does not have an example that have more than one columns as the grouping key in functions.py