Automatically filter out barcodes with too many molecules

Currently the filtering is based on the number of molecules in the DBS cluster and on an average size of 30kbp. Instead we could take into account the actual size of the molecules and how much of the genome that is covered by the DBS cluster. Then we could set a probability threshold or similar for which clusters to remove/mask.  

Issue copied from: https://github.com/FrickTobias/BLR/issues/78