histo slowdown

[caseywdunn]

In this commit a4b6dfd I changed from vectors to hashmaps for compiling the histograms, so I could get all values and validate number of kmers. It led to a slowdown from about 30s to 90s for Consolidating chunks and creating histograms.

Try nohash for speedup. https://stackoverflow.com/questions/70551997/faster-hashmap-for-sequential-keys

If that fails maybe go back to vectors.

[caseywdunn]

More example of nohash hasher - https://nucleistudio.github.io/governance-os/nohash_hasher/struct.NoHashHasher.html

[caseywdunn]

Encapsulating data structure for Histogram inside new Histogram structure leads to slight further slow down to 101s. But now I can change internals in one place without needing a bunch of type changes all over the code.

[caseywdunn]

Now takes 38s with IntMap instead of HashMap. So back to almost the same speed as using a vector. Wow!