Funannotate

* added support to `funannotate predict` for an `--other_gff` option that will pass annotation directly to EVM. You can control the weight for EVM, like this `--other_gff my_predictions.gff3:10`, which would give the gene models a weight of 10 in EVM
* better support for `--pasa_gff` passed to `funannotate predict` where now input is not hardcoded to have `transdecoder` in column 2 of the gff file. You can also control the EVM weight like this: `--pasa_gff my_pasa.gff:10` to give it a weight of 10
* BRAKER1 method now pulls out high quality Augustus models (HiQ) that have >90% exon supported by evidence, these are given a weight of 5 in EVM
* Added a few stats for repeat masking genome as well as number of transcripts mapped
* updated funannotate so it is compatible with new version of GAG v2.01.

* Numerous bug fixes
* Strip asterisks from protein fasta files to avoid problems with InterProScan
* logfiles folder was not being created if `--genbank` was passed to `funannotate annotate`
* Linux bug where last step of `funannotate predict` was terminating prematurely resulting in partial output files

* Re-write of the InterProscan parsing scripts. Now script will parse IPR Domains and GO terms directly from XML file, instead of splitting XML file and then parsing 1 by 1.
* Great update by John Longinotto on his [pybam](https://github.com/JohnLonginotto/pybam) native BAM parser which is integrated into `funannotate predict` to quickly check BAM headers to make sure they match FASTA headers for input into Braker

* fix tRNA gene model filtering to deal with the `tbl2asn` >150 error
* improve XML parsing in `funannotate compare`
* add `diamond` alternative for `exonerate` pre-filtering in `funannotate predict`
* make `funannotate` docker compatible and create docker image
* EggNog and BUSCO2 database are now not downloaded in the initial setup, but you can manage EggNog databases with `funannotate eggnog` . This was due to problems in building docker image downloaded the large databases. The scripts will download on the fly if default database is not available.
* added some external dependency versions in `funannotate check`

- bug fixes for logging
- bug fix when multiple protein evidence files are passed
- add phobius to funannotate annotate to predict secreted proteins in combination with signalp
- add test data `genome4.fasta` that can be used to test the BUSCO2 augustus training method
- added support for checking BAM reference sequence headers if they match the genome FASTA headers, this only happens if BAM file passed to `--rna_bam`

- typo fixes for log file names
- typo fix for fuNOG annotations in secondary metabolism module, this was now fixed to use the proper `--eggnog_db`option
- test for dN/dS ratio test to assert that the tree that was drawn by Phyml has the correct number of proteins
- new feature for BUSCO models if `--ploidy` is greater than 1 in `funannotate predict` that duplicated BUSCO models are also parsed, the one that is picked has the highest score
- Support for bypassing RepeatModeler/RepeatMasker, you can now enter `--masked_genome` and `--repeatmasker_gff3` options to skip that step. Note that both options are required.

- update to tblastn/exonerate protein mapping for better speed and more thorough searches
- added `--ploidy` option to `funannotate predict` which controls the max number of hits for tblastn filter to pass to exonerate, which is set at 2 x ploidy. You should likely only increase this if your assembly is more than haploid - so perhaps newer assemblies with nanopore/pacbio may be able to resolve diploid chromosomes. It shouldn't have negative consequences in increase `--ploidy`, but will increase run time for protein mapping.
- happy holidays...