Proposed revision to lump functionality

[nsheff]

I propose we revise the lump arguments to these:

• --lump-n <n>: (currently lumpn) puts n samples together
• --lump-s <size>: (currently lump) puts together samples until they reach size in total input file size
• --lump-j <jobs>: (not currently implemented) divides samples into jobs different jobs

This adds the lump-j functionality (which should be really easy to implement, just piggyback on the lump-n functionality). And then, it streamlines the terminology; I'm always confused at whether I should use lump or lumpn.

Thoughts:

• this would be a (minor) change to the UI
• is the new lump-j useful enough? Only if implementing it is very simple (which it should be)
• are the names intuitive? -s, -j, and -n? I think -s could mean either size or sample. But I can't think of anything better.

[donaldcampbelljr]

Better to think of it as starting with number of jobs and then having looper distribute samples across jobs (processors).