WIP: Update to automatically work with our current pyAFQ outputs

[arokem]

We've gotten some reports that this doesn't currently work smoothly (e.g., in yeatmanlab/AFQ-Browser#292), so I am going to use this branch to investigate and resolve.

[arokem]

Some notes for myself before starting: I think that what I want to do here is to make it so that if a BIDS dataset has a participants.tsv file and an afq derivatives folder then afqbrowser-assemble will use this as the nodes.csv and subjects.csv files mentioned in our current documentation. I believe that we have code in pyAFQ that will create a streamlines.json file from the first subject in a dataset. But we might want to let users also choose the subject that will be used as the exemplar. I don't think that we want to support selection of any subject through the UI, but I will look into that.