Explicitly specify histogram bin endpoints/widths

George has convinced me that we should explicitly specify the bin labels of our histograms rather than rely on a formula and lots of special cases for short histogram lengths.

The most straight-forward thing is to let researchers specify a list of bin labels.. It might be more compact to encode if this list is a list of widths, but that may just be an encoding decision for the consensus. We should probably choose simplest possible representation here.