Optimize bc_demultiplex

[jarondl]

Embarrassingly, bc_demultiplex can be the most time-consuming step in a pipeline.
We should profile it to check what are the causes for this, and optimize. Ideally, we should benchmark this on real-life data.

After profiling, possible solutions could be:

• Simple optimizations of inner loops in Python. Sometimes even assigning local variables can make a difference.
• Stronger optimizations of inner loops - maybe use a C-based package, or Cython
• Parallelize.

@flo-compbio , @gufranca