Bug in initializing founder genome count.

[molpopgen]

When a DiploidPopulation is constructed, it is passing N to the base class. This base class knows nothing about ploidy, resulting in the founder genome having a count of N and not 2N. This was discovered quite indirectly when working on #828 and was only possible to catch due to molpopgen/fwdpp#353

While working on #828, unexpected runtime errors were getting raised. I've tracked it down to the add mutation function, but the exact cause is not known. The symptom is that a diploid can end up referring to an extinct genome.