Implement repooling functionality

We have the ability to go back and resequence samples that didn't yield enough reads the first time. This could and should be automated with the database. Conceptually, this will require:

1. Parsing demultiplexing stats output (I am working on some code for this, but it's not tested yet)
2. Making a decision on which samples to repool and at what quantity
3. Creating a pick list for these samples
4. Creating a valid sample sheet for the repool

#4 will require a keeping a couple things in mind. First, the possibility of sample name collisions—currently sample names only must be unique on a plate (or maybe in a study). Repooling could violate this. What to do? Second, the possibility of barcode collisions. We only *need* to enforce barcode combo uniqueness per lane. Repooling samples from multiple lanes raises possibility of violating this. Need to be aware and come up with solution.