Randomize reads before quantification with Salmon

### Context

Salmon assumes that the reads you give it are in a random order -- see the first **Note** [here](https://salmon.readthedocs.io/en/latest/salmon.html#using-salmon) in the Salmon docs.

### Problem or idea

Because we won't know if the `fastq.gz` files from SRA have been ordered in some way without checking, we might want to shuffle the reads before giving it to `salmon quant`. It is not clear to me how frequently these files will be in a non-random order (if ever).

### Solution or next step

Based on my understanding of #157 (and by extension, how things are currently implemented), we have to unzip the `fastq.gz` file to determine the read lengths. That might be the time to shuffle the reads, but this needs a dev team perspective. Also, as noted above, I don't know how many samples this will effect.

Review issue #157 and the Salmon documentation note about read ordering first. Trace where fastq.gz files are prepared and passed to salmon quant, then determine whether shuffling belongs in that path; done means the team has an agreed implementation point and verified behavior for affected samples.

Written by the indexing model from the issue text.