Output un-normalized coverages from "dynam covskew"

[fedarko]

Useful to provide context on how high-coverage a contig is, relative to other contigs; also, more importantly, useful if normalized-coverage-clamping happens for min / max skew bins. In this case, it would probably make sense to use unnormalized coverages for computing the PTRs?

Should also update the tutorial about this.

• Output unnormalized coverages (i.e. just the median coverage in each bin). The easiest way to handle this is just adding an extra column to the TSV output of covskew called MedianCoverage or something.
• Test
• Update tutorial to explain
• In tutorial, show that PTRs produced using unnorm and norm coverages are roughly the same (assuming that clamping has not happened, or at least hasn't impacted the min/max skew bins' norm coverages)