Random barcode choices in correct.py

[NickSto]

As of version 2.16, correct.py often chooses different barcodes to use than in the regression test output. It's stochastic, though. There are at least 4 different outputs it randomly produces, all with different barcode choices.

This issue does not appear to be present in 2.15.

Note to self: These are the files correct.bug.*.tsv in tests.