Support "partition by XXX order by XXX" SQL

Based on our expereince, looks like for bounded stream, beam SQL does not able to support statement like "partition by XXX order by XXX". It will not be able to parition data to different nodes and sorting data in each partition/node parallelly.

We have to use Java SDK and extension to convert following SQL statement to GroupByKey **** SortValues to achieve this. 

Does we miss anything? If not, is this something that we can improve? and took a quick look at calcite, seems that it can explain the query plan for "partition by... order by..." fine.

Imported from Jira [BEAM-6105](https://issues.apache.org/jira/browse/BEAM-6105). Original Jira may contain additional context.
Reported by: BrandonJ.