Enable automatic dataset splitting

[SebieF]

In addition to providing the dataset splits via the input FASTA files, it would be a nice feature to allow automatic dataset splitting within the biotrainer pipeline. The dataset splitting should not be done based on pure random splits, but should also feature checks for sequence similarity within the dataset splits, using mmseqs.

Steps include:

• Create new config file option(s) for automatic dataset splitting
• Add a step in the pipeline where the splitting is done
• Run mmseqs and split afterwards
• Provide feedback in the out.yml file and the logging about the splits