Support Grouping Sets

[henneberger]

This enables more advanced types of aggregations like ROLL UP and CUBE:
https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sql/queries/group-agg/

Extends support in the LPRewriter for grouping sets. In particular, add a separate primary key column as the md5 of all group-by columns if a) one of the columns can be null or b) there are multiple grouping sets. Otherwise, keep the primary key as it currently is.