Dnaapler

* Thanks to the one and only [rrwick](https://github.com/rrwick), Dnaapler now supports the [GFA](https://gfa-spec.github.io/GFA-spec/GFA1.html) format as input. This was done to ensure support for Ryan's new bacterial genome assembly tool [Autocycler](https://github.com/rrwick/Autocycler), the successor to Trycycler, but may also be useful if you have GFA files from e.g. Unicycler, Flye, Spades or other assemblers.
* If you run `dnaapler` with GFA input, you will get a GFA output as well.
* If you run `dnaapler` with GFA input, only circular contigs will be reoriented
* Relaxes the `MMSeqs2` dependency to `>=13.45111`

* Adds support for reorienting contigs where the gene of interest spands the contig ends - [fixes this issue](https://github.com/gbouras13/dnaapler/issues/90). Thanks marade oschwengers.
* Specifically, this is done by rotating each contig in the input by half the genome length, then running `MMseqs2` for both the original and rotated contigs. The `MMseqs2` hit with the highest bitscore across the original and rotated contigs will be chosen as the top hit to rotate by, therefor enabling detection of partial hits (on the original contig) that span the contig ends.
* This has only been implemented for `dnaapler all` (this should be the command used by 99% of users).

* Thanks to the inimitable [rrwick](https://github.com/rrwick), v1.0.1 is a patch fixing a string-parsing bug.
* If your contig headers were integers, `dnaapler` did not rotate the found `BLAST/MMseqs2` hits. This was a pre-existing issue (not introduced by v1.0.0).

* **BREAKING CHANGE** - `dnaapler` now uses `MMSeqs2` rather than `BLAST`. You will need to install `MMSeqs` if you upgrade (if you use conda, it should be handled for you)
* There are 2 reasons for this:
1. Users reported problems installing BLAST on MacOS with Apple Silicon (see e.g. [here](https://github.com/gbouras13/pharokka/issues/368)). MMseqs works on all platforms and is dilligently maintained.
2. MMSeqs2 is much much faster than BLAST (what took BLAST a few minutes takes MMSeqs2 seconds). We should have written `dnaapler` with `MMseqs2` to begin with.
* The alignment resuls may not be identicial (i.e. they might find specifically different top hits), but the actual reorientation is likely to be identical (at least in my tests). Please reach out or make an issue if you notice any discrepancies.

For example - on my machine (Ubuntu 20.04, Intel i9 13th gen 13900 CPU with 32 threads), for a _Staphylococcus aureus_ genome with 1 small plasmid, `dnaapler -i staph.fasta -o staph_dnaapler -t 8` took ~129 seconds wallclock with `v0.8.1` using `BLAST`, while it took ~3 seconds wallclock with `v1.0.0` using `MMseqs2`.

* Minor release - adds `--db dnaa,repa,cog1474` as an option for `dnaapler all` to allow for archaea orientation in hybracter

* Adds `dnaapler archaea` and adds archaeal reorientation functionality into `dnaapler all`
* Specifically, this uses 403 COG1474 genes [COG1474](https://www.ncbi.nlm.nih.gov/research/cog/cog/COG1474/)
* Relaxes (to warning) where no BLAST hits are found - pipleine will still complete (requested in a number of issues 74 76 77)