Funannotate

* add `funannotate test` as a unit test script to validate installation 212
* add `CodingQuarry` and `StringTie` integration into funannotate. Note these are "silent" dependencies, meaning if not installed this method will be skipped. If both tools are installed and RNA-seq data is used they will be run automatically 200
* fix contig number reporting in `funannotate clean` 210
* fix bug in a few of the accessory tools `funannotate util`
* update database to MiBIG v1.4

* fix bug in `train` and `update` where script would die due to symlink error 189
* for `predict` the `--protein_alignments` option now takes GFF3 input (not exonerate output). This is to make consistent with the `--transcript_alignments`. Scripts now write hints file for augustus from the GFF3 file.
* similar to above, added `funannotate util prot2genome` which will run the diamond/exonerate mapping of proteins to the genome --> output is GFF3 file compatible with EVM. These data can then be passed to `--protein_alignments`

* bug fix for `funannotate predict` during parsing the soft masked genome -- for large genomes this was slow and used too much memory. it is now multithreaded and has lower memory footprint. 197
* bug fix for `ncRNA` models are now listed as full length, should no longer cause NCBI errors 195
* support multiple inputs to `--other_gff` 191
* make `augustus` use `--softmasking=1` option
* default value for `--soft_mask` is now set to 2 kb `2000`
* output fasta files are now wrapped at 80 characters
* `tbl2asn` is now multithreaded on large genomes or those with more than 10000 contigs
* several updates to parsing of GenBank files to deal with unexpected formatting 196

* support for long-read RNA-seq data: `funannotate train` and `funannotate update` can take PacBio isoSeq (`--pacbio_isoseq`), Nanopore cDNA reads (`--nanopore_cdna`), and Nanopore direct mRNA (`--nanopore_mrna`).
* fix for important bug in transcript alignments in `funannotate predict` -- bug in previous versions related to multi-exon crick alignments not getting correctly parsed into GFF3 alignments
* soft masking is now decoupled from `funannotate predict`, this is now done with `funannotate mask`. Reason for this switch is to allow more flexibility in how the assembly is soft masked -- this can be done externally with another program. This change will allow users that don't have access to RepBase to use an alternative from RepeatMasker/RepeatModeler. One alternative is RED -- I wrote a wrapper for called [RedMask](https://github.com/nextgenusfs/redmask)
* `funannotate predict` can now run without GeneMark being installed -- again to accommodate users that may be unable to use GeneMark due to licensing. Note you can pass gene predictions from any external program to `--other_gff` and they will be handed off to Evidence Modeler.
* spaces in either strain or isolate name will be stripped 180
* default program for `funannotate clean` changed to minimap2 176
* fix errors in partial gene models derived from using EVM script to generate proteins, this is now done internally using exact coordinates 184
* added `--soft_mask` option to `funannotate predict` which will control the option with same name in GeneMark, i.e. default is `--soft_mask 5000` which means that repeat regions less than 5 kb will be ignored for GeneMark prediction, those greater than 5 kb will be fed to Genemark. 185
* bug fixes for `tbl` file generation. all tRNA models will be partial 184
* improvement to how data from `funannotate train` is used in prediction steps
* Slight changes for clarity to `funannotate predict` flags for evidence alignments:

--protein_evidence Proteins to map to genome (prot1.fa prot2.fa uniprot.fa). Default: uniprot.fa
--protein_alignments Pre-computed exonerate protein alignments (see docs for format)
--transcript_evidence mRNA/ESTs to align to genome (trans1.fa ests.fa trinity.fa). Default: none
--transcript_alignments Pre-computed transcript alignments in GFF3 format

* added `funannotate util bam2gff3` script to convert coordsorted RNA-seq BAM alignments to GFF3 compatible alignment file.
* fix bug for input of files+weight in `funannotate predict` -- script would get hung up if you passed `--other_gff snap_alignemnts.gff3:5` 191
* allow for non-standard LocusTags - will now split on last underscore 191

* bug fixes for sec met cluster output files and corresponding MiBIG cluster mapping
* add `tRNAscan-SE` to `funannotate check` and `predict`
* update menu with some params that were missing

* bug fix for `funannotate compare` where GO enrichment not being run in parallel from last update
* use diamond blastp search for ortholog detection --> speed increased.
* don't run seqclean if file present
* update docker release to newest version of funannotate as well as newest version of Trinity, PASA