Funannotate

Bug fix release, recommended for all users.

* use `uuid` unique identifier for tmp file names (previous uses process id)
* fix `diamond` command in mapping proteins to genome 529
* add some contig name checking for inputs to predict -- some users erroneously pass GFF3 files that reference transcripts or some other assembly, warn user 528
* add support for antiSMASH v6 531 539
* check genome for IUPAC errors 532
* re-write long reads mapping to trinity transcripts 539
* allow for fewer busco models for training 556
* expose some EVM parameters to predict 558
* modify EVM partitioning -- some gene models close to partition were not being called correctly 558.
* support newly released eggnog-mapper >= v2.0.5 566, note that v2.x - 2.0.4 will not be supported as the output is broken in the sense that it is impossible to parse the proper EggNog reference ID. If you are using any eggnog results from this code, strongly advised to re-run with new release.
* add `_pasa` to PASA database names 398. Note -- this change could potentially cause pre-existing databases run with older funannotate code to not be recognized properly after v1.8.5.

Bug fix release
* remove auto fix for RNA-seq data if it does not have Trinity compatible headers -- warn user to fix instead 485
* fix antiSMASH v5 parsing 490, update secmet GBK format of clusters
* add `funannotate util gff-rename` script
* use `natsort` for sorting contigs/names so gene names are correctly auto incremented 501
* fix for signalp v5 494
* support fasta headers with spaces for clean 504
* rstrip(*) from proteins if GFF/FASTA passed; 508
* expose --p2g_prefilter to use tblastn for mapping protein evidence 495
* add --anysymbol to mafft calls 514
* use frameshift diamond if >2.0.5 503
* improve glimmerhmm parsing error report 519
* add `--no-progress` option to clean up terminal output (ie log files from cluster)
* use sorted bedtools intersection to reduce memory usage 522
* add --trnascan option to predict 523

* highly recommend all users upgrade
* Code now python 2/3 compatible (only 9 months late....)
* EVM now uses `interlap` for slicing input files, significant speed increase for larger genomes
* if `--organism other` is passed, `codingquarry` is now disabled by default -- you can turn it on by specifying a valid weight, ie `--weights codingquarry:2`
* support for `signalp` version 5
* move resources to JSON download via GitHub, will allow for updating/changing resources without re-installing a new version of `funannotate`
* several fixes for more robust GFF3 parsing
* fix for cazyme assignments where results were duplicated; also upgraded to database version 8
* fix bug where PASA results not properly passed to EVM if funannotate predict was re-run
* many other bug fixes

Okay, so apparently I lied about bug fixes before py3 release. Here are some quick ones.

Also -- some users have said that conda recipe fails to find a solution, with the help of reslp this is perhaps caused by `ete3` package as a dependency. Since `ete3` is only used in `funannotate` for the dN/dS calculation in `funannotate compare`, I'm going to remove it from conda recipe as a dependency. Users that would like to continue using the dN/dS function in compare will need to manually install ete3 (ie conda could work on your system or pip, etc).

* update the log file copy/remove in `funannotate annotate` which causes a problem in singularity container
* encoding bug fixes for some PFAM results
* fix for `funannotate iprscan` which was not correctly combining XML files with newer versions of Interproscan.

* bug fix release. This will be the last release before updating the code for py3
* bug fix for remote 379
* improve GFF3 parsing in `funannotate annotate` allow for non-funannotate identifiers (hopefully this catches most of them)
* `--aligners` option was missing from help menu in `train/update`
* add `--debug` flag to `funannotate iprscan` -- there is potentially still an issue here combining results from the newest release of Interproscan -- I need some intermediate results to see what the problem is
* fix GO annotations parsing for more versions of goatools 363
* fix for bug in the "gene start or end in a gap" function
* make sure `funannotate annotate` re-runs tbl file generation if script is re-run
* fix in PASA function logic where PASA db not found if using --gff and --fasta input.

* busco internal run now uses the local augustus config path, so hopefully write access to `$AUGUSTUS_CONFIG_PATH` is no longer required
* long reads are now processed to remove forward slashes from names if present -- this was causing problems with PASA 326 250
* added augustus hints file generation when using RNA-seq data. Also added a check for `--transcript_evidence` if found in `--transcript_alignments`, ie so extra transcripts can be added instead of only those used to generate alignments in `funannotate train`. 360
* fix typo in `funannotate check`
* fix for proteinortho issue in `funannotate compare` 350