Better documentation for smck with continuous genome.

[jeromekelleher]

In #2445 we added documentation for the SMCk which says

We recommend
SmcKApproxCoalescent(1)`` for discrete genomes instead, or
 SmcKApproxCoalescent(1e-14) for continous genomes

This isn't sufficient, we need to give some explanation for where this number comes from and why it's different when you've got a discrete gemome.