Memory savings in the Tesserae algorithm #6

From #5 (review)

Are there still some useful memory reductions to be obtained by storing only certain columns of the traceback and computing forward when necessary? Ultimately my goal will be to treat ultra-long Nanopore reads (~0.1 Mb, maximum ~3 Mbp) with Tesserae. But I'm not sure what the engineering requirements are to get there yet.