[feat] Feat/hash algs

[St4NNi]

Draft for hash algorithms

This PR adds different hash algorithms to the sketch command. This includes:
xxhash, ahash and legacy murmur3.

Ahash uses the ahash fallback algorithm and is only suitable for k < 32 since it internaly relies on a u64 bit-kmer representation. The raw hashing performance of ahash is 10-30x faster compared to existing hashing algorithms [0.3ns vs 3-8ns per iteration!]
The PR will be merged as soon as all new methods are sufficiently tested (and documented).
Caveat: This currently needs rust nightly to build properly, the final release will try to remove this constraint.

Changes

• Implemented lots of new features to sketch
• Added sourmash compatible outputs (import other kinds of fracminhash etc? sourmash-bio/sourmash#2710)
• Added different hashing algorithms
• Added bit-kmer support for k < 32 and regular canonical kmers for larger kmers
• Update README.md
• Added basic tests for hashing performance and bit-distributions including a crude Kolmogorov–Smirnov test to test against equal distribution

Sourmash outputs

You can use jam-rs to create sourmash compatible outputs, a basic example looks like this:

$ jam sketch <INPUT> -o <OUTPUT> -k 50 --fscale <SCALE> --format sourmash --algorithm [default, ahash, xxhash, murmur3]

Please be aware that this will produce differing results for k < 32 if you compare the output to sourmash generated outputs, since this internally uses a bit-kmer encoding before hashing instead of regular ASCII encoding.

• default algorithms means ahash for k < 32 and xxhash for larger kmers
• fscale is the compatible parameter to sourmashs scale

So:

$ sourmash sketch dna -p k=50,scaled=1000 <INPUT> -o <OUTPUT>

and

$ jam sketch <INPUT> --fscale 1000 -k 50 --format sourmash --algorithm murmur3 -o <OUTPUT>

will produce similar results.

Feedback

Please feel free to give feedback for additional updates / changes. This is only the first step to improve sketching functionality the improvement of search especially for smaller fragmented sequences in larger sets will come next.

[codecov-commenter]

Welcome to Codecov 🎉

Once merged to your default branch, Codecov will compare your coverage reports and display the results in this comment.

Thanks for integrating Codecov - We've got you covered ☂️

[St4NNi]

This is still WIP, but I have decided to upgrade on this incrementally to keep PRs reviewable !
Stay tuned more updates will follow soon !