Funannotate

- bug fix for `funannotate compare` where the genome stats was not printing for all genomes
- goatools changed their headers on the output of the GO enrichment (again), so re-wrote how the data is parsed, hopefully this fix applies to all versions.

- explicitly run `rmblast/ncbi` engine for RepeatMasker to avoid problems if user has default setup as something else, i.e. DFAM. Note you still need to install RepBase Libraries, e.g.


wget --user name --password pass http://www.girinst.org/server/RepBase/protected/repeatmaskerlibraries/repeatmaskerlibraries-20150807.tar.gz
tar zxvf repeatmaskerlibraries-20150807.tar.gz -C {HOMEBREW_PREFIX}/opt/repeatmasker/libexec

cd {HOMEBREW_PREFIX}/opt/repeatmasker/libexec
./configure <config.txt

- bug fix to `funannotate compare` that was not pulling orthology groups for the final annotation table
- added `transcription factors` to output of all annotation table.

- bug fix in `funannotate compare` that was calculating MEROPS summary stats incorrectly
- added `--minlen` option to `funannotate sort` to discard short contigs

- move install to the funannotate wrapper, `funannotate setup`
- fix bug with custom input folders in `funannotate annotate`
- output proteins/transcript files for both `funannotate predict` and `funannotate annotate`

- bug fix in `funannotate predict` when using `BUSCO` the EVM input was pulling entire gene models instead of sliced models
- bug fix where `BUSCO` models were within 100 bp of the start or end of contig resulting in a `bedtools` range slicing error
- remove slicing of hints file for parallel AUGUSTUS method as splitting the hints file was a slow, faster to just pass the entire hints model to each contig chunk and let AUGUSTUS filter it.