Fix for abundance deviation in metagenome mode

[lcoombe]

• If simulated read longer than selected chromosome, select from same species preferentially
• Previous behaviour randomly selected a different sequence from the full pool of species
• In the example of the Zymo mock model, this meant that S. aureus reads ended up under-represented, while Cryptococcus was over-represented
  • This was because the S. aureus reference contained 4 sequences - the main circular genome and 3 short plasmids
  • When a sequence from S. aureus was randomly selected, frequently the plasmids were chosen, which were shorter than the requested read length
  • Because there are more Cryptococcus sequences in the reference compared with the number of sequences in other references, randomly choosing a replacement sequence from the entire pool of species/sequences meant that Cryptococcus was chosen more frequently
• To retain the requested abundances as much as possible, preferentially choose the 'alternative' sequence from the same species
  • This is consistent with the version used in the meta-NanoSim paper
• As a fall-back, if there are no appropriate sequences in the species' reference, choose another species
  • If this is required, a warning will be printed, advising the user to check the abundances after simulation finishes