Using a known pedigree to assess sequencing error

[hyanwong]

We've been saying that there's no obvious way to assess the degree of sequencing error in real data, and therefore to know the best settings for the mismatch ratio in tsinfer. But if we have a known pedigree, then we can probably use that to estimate the amount of sequencing error in our dataset. Starting on this would make a nice student/MSc project I reckon.