Use GNU-parallel to run prodigal faster - #267

@mruehlemann the problem with this is that you do not want to do it when using isolate genomes because the gene models will be much worse when trained on only 100kbp of sequence.

for metagenome it does make sense, but you probably would want to ensure every chunk had some big contigs to ensure the training was good. even better is to chunk by MAGs. another option is to collect the biggest contigs, train prodigal on those, then use that model (in parallel like you did) on the chunks.

I think these are ideas I plan to put in mokka https://github.com/tseemann/mokka