Add a _simulate() method that directly uses msprime as the backend so we can get the haplotype data

[isaacovercast]

• Provide a function to get the actual ms output ** The simulated data is kind of buried in the momi code, and would actually be a pain to extract.