Allow users to configure partitioning for pipe APIs

[fnothaft]

I'd like users to be able to configure the partitioning for piped commands. Specifically, I'd like to be able to choose between a few strategies, such as:

• Default (our current strategy, where we partition by contiguous genomic region)
• Partition by target (allows users to provide a second GenomicRDD, and does something like a regionJoinAndGroupByLeft using the second RDD) --> This would be useful for things like calling variants on an exome panel
• Skip partitioning (what happens if no sequence dictionary is attached)

Thoughts/comments?

[heuermh]

From what I see, #1758 fits into this use case.

[heuermh]

Closing as WontFix