Mikado

Fix [404](https://github.com/EI-CoreBioinformatics/mikado/issues/404): Error in extract_promoter_regions.py helper script

Fix [405](https://github.com/EI-CoreBioinformatics/mikado/issues/405): Error in remove_utrs.py helper script

Also fixes a name clashing issue in daijin's configuration object and other small bugs

Fixes class2 for the Daijin pipeline and removes a noisy print in BED12 parsing

Fix [401](https://github.com/EI-CoreBioinformatics/mikado/issues/401): Cannot run Class2 in Daijin pipeline

Fix the Github Action tests, moving to use only Python 3.7+ (due to AsyncIO), and updated the documentation.

Fix [385](https://github.com/EI-CoreBioinformatics/mikado/issues/385): clarified and corrected the tutorial for Daijin.
Fix [389](https://github.com/EI-CoreBioinformatics/mikado/issues/389): Mikado now can theoretically output BED12 files from Mikado prepare. This is still
Fix [395](https://github.com/EI-CoreBioinformatics/mikado/issues/395): corrected an issue in mikado prepare that left models with incorrect proteins in the input annotations. This caused mikado pick to crash during the padding procedure.
Fix [396](https://github.com/EI-CoreBioinformatics/mikado/issues/396): corrected and clarified the errors related to missing configuration files (e.g. incorrect scoring files being provided).
Fix [397](https://github.com/EI-CoreBioinformatics/mikado/issues/397): corrected a small bug in mikado prepare, when providing the input files through the CLI rather than through the configuration file.

Testing Mikado also on OSX, and adding OSX to the supported OSes in Conda.

Fix [392](https://github.com/EI-CoreBioinformatics/mikado/issues/392): Mikado was having trouble with pipes in the sequence IDs (either present in the first place or added by NCBI+ when using `makeblastdb -parse_seqids`).

Added the SHM capabilities to `mikado serialise` and `mikado compare`.
Now `mikado` `serialise`, `pick` and `compare` will copy the database to the TMPDIR before running, to ensure atomicity.
This will also prevent problems in case the data is usually on an NFS drive, as is the case in most computing clusters.

Pinning sqlalchemy to <1.4.0 until sqlalchemy_utils is updated.
Moreover, solved a small bug in `prepare`: setting `prepare.exclude_redundant` to `True` in the configuration file had
no effect. Now it is equivalent to use the `-er` switch on the CLI.