Allow mutations only at integer sites

[petrelharp]

I've written code that - I think - puts down mutations only at integer positions. With this branch, it worksl like this:

import msprime
import pyslim

ts = msprime.simulate(10, recombination_rate=0.01, length=100, mutation_rate=0.1)
new_ts = pyslim.mutate(ts, mutation_rate=0.2)

for h in new_ts.haplotypes():
    print(h)

It's simulating under an infinite alleles model, to avoid having to check if the parent mutation is the same. We want this in tskit eventually, and implemented in C (this is pure python, so probably slow for big tree sequences) but we could provide this as a stopgap.

[jeromekelleher]

One quick question here: why the mutation_rate=0.1 in the call to msprime.simulate()?

[petrelharp]

To check if existing mutations would be retained... which, I don't think they are, correctly. See tskit-dev/msprime#553