Stale reference and reads

[jeffreybarrick]

When breseq is re-run when it already has an existing output directory but the user has changed the input files, it is not intelligent enough to erase the relevant intermediate results and start the anaysis over. This can lead to confusion.

Ideally, breseq would make a quick hash of the input files to verify that they are the same and delete the intermediates if they have changed. Since read files can be quite large, the downside of this approach is that calculating the hash will slow down normal runs a bit.

Possibly use this for hashing files...
https://github.com/RedSpah/xxhash_cpp