[Meta] Memory and runtime performance improvements

[standage]

Here I want to keep track of some ideas we've thought of and, in some cases, discussed to improve kevlar's memory consumption and speed. I'll link to relevant issue threads and pull requests as they come about.

• Buffer abundance queries in kevlar novel (see Buffer abundance queries in kevlar novel #208, Improve performance by caching k-mer abundance query results #276)
• Use a faster hash function (see Replace murmurhash with cyclic hash #209)
• Re-visit and re-evaluate error correction (Re-evaluate impact of error correction #273)
• Populate dedicated counttables for the kevlar simlike step, to enable higher FPR for kevlar novel step (Populate counttables specifically for kevlar simlike step #274)
• Investigate sketch composition so controls can be combined for kevlar novel step (Compose control counttables for kevlar novel step #275).

[standage]

It turns out that applying k-mer filtering mask at k-mer counting time (kevlar count) has all kinds of downstream efficiency benefits. See #277.