Scoary

- Another bug fix related to the restrict_to option where Scoary would crash if it was NOT set.

- Bug fix. The --restrict_to option had become broken in 1.2.0 because one function was passing the full set of isolate names rather than the restricted set downstream in the analyses. This has been fixed.

- Major changes. Scoary now implements the pairwise comparisons algorithm to account for population structure.
- Scoary now imports 4 classes: Matrix, QuadTree, PhyloTree and Tip. They two former are used for storing pairwise distances between isolates, and the two latter are used for the pairwise comparisons algorithm. Scoary_methods contain some new functions and most of the old ones have been altered to allow the pairwise comparisons implementation. However, there are no changes to how Scoary calculates the previously implemented statistics.
- There should now be a significant speed increase, as Scoary now stores Fisher's exact test p-values rather than re-calculating for every gene. Previously, a new calculation was made for every gene, even when the 2x2 table was identical.

- Fixed another bug related to Benjamini-Hochberg p-value adjustments. (Thanks again to cimendes). The numbers should now correspond to R's "p.adjust" method IF the number of tests are the same. (Scoary runs the correction on ALL genes, not just those with p<.05)
- Results are now printed with "" quotation marks around each cell to avoid weird cell breaks if annotations contain semicolons when opening in spreadsheets.

- Fixed a bug where adjusted p-values were not being calculated and output correctly. (Thanks to cimendes).
- Benjamini-Hochberg p-values are now calculated from least to most significant (step-up) instead of step-down.
- Holm-Sidak p-values has been removed.
- The Scoary version is now displayed in the help message and if calling Scoary with the --version flag. (Thanks to EricDeveaud)

- Scoary now imports all methods from Scoary_methods in order to circumvent errors when trying to run Scoary under Python 3.x.
- Genes that have the same naïve p-value now have adjusted p-values penalized with the same factor rather than an increasing one.