Add support for error profiles

[bjclavijo]

Error profiles, given by name to the sequencing steps, should change how errors are distributed in the reads.

At the very least, besides the current profiles (i.e. "uniform"), we should have a "degrading" profile for paired reads (i.e. where the ends of the reads have more errors than the beginning and r2 has more errors than r1), and a "homoindels" (i.e. where mostly homopolymer lengths are changed) and a "homoindels_clustered" (i.e. same as homoindels, but also having a nearby error makes it more likely to have errors in the region, so reads have "high" and "low" quality sections).