Implement predicate pushdown for GROUP_BY Spark SQL

[adersberger]

Cool showcase:

select metric, process 
from timeseries 
group by metric, process
order by timeseries

With and without GROUP_BY pushdown and ORDERING in Spark