documentation and support for other implementations of MinHash and FracMinHash

[ctb]

revisiting a (5 yro :) conversation about interoperability marbl/Mash#27 after some discussions on slack with @kescobo on µbioinfo slack. See also a motivating question about implementations in BioJulia BioJulia/BioSequences.jl#243 and this issue in rkmh edawson/rkmh#3.

Here in sourmash, I think we could do more to support alternative implementations of MinHash and FracMinHash in other projects. I'll update this issue more as I think through them, but off the top of my head -

• provide both very simple and more complicated example sketches, sequence files, and simple Python code to run them all
• provide Jaccard (MinHash) and Jaccard, containment x 2, avg containment, max containment, and angular similarity (FracMinHash) numbers, as well as providing the underlying equations
• describe the current sourmash JSON format and provide simple example import/export scripts for sourmash in Python and R (since we already have them ;)
• talk about converting sourmash sketches from (e.g.) wort into regular minhash, where possible. (that is, sourmash sig downsample)
• show how to download, export, make use of some SRA sketches

The goal is to make it easy for others to implement simple test suites to check on interoperability of the actual sketching and sketch comparison code.

[ctb]

on slack, luiz pointed out that we already have good Rust interoperability b/c, well, if you scratch sourmash's Python layer just a little bit, you find Rust underneath. And:

There is a C header for doing the interaction between Rust and Python, here: https://github.com/sourmash-bio/sourmash/blob/latest/include/sourmash.h

[ctb]

also: we should provide some simple example code for making use of some of our pre-computed sketches from wort, and converting them into mash-usable sketches.