Compute oxidation

[luizirber]

This PR expose functionality in a way that doesn't requiring moving all the Python compute code into Rust.

• I moved the internal functions from compute (make_minhashes, add_seq, build_siglist, save_siglist) to the outside scope, and made all the arguments explicit. It would be hard to refactor otherwise...
• add_sequence and add_protein methods for SourmashSignature. The idea here is to only cross the FFI layer once for each sequence, and being able to control better what can be improved on the Rust side (like using rayon for parallelization?).
• add_protein in MinHash was also moved to Rust, and it is 100x faster 🤣 /cc @olgabot
• Benchmarks in Python (for add_protein) and in Rust (for add_sequence)
• add_sequence in Rust is way simpler, and way faster. Turns out doing simpler things and letting the compiler optimize was better than my micro-optimizations =P
  • but there is still a trick: I pre-calculate the reverse complement for the full sequence, and then while moving the k-mer window forward for the sequence I move another one backwards for the RC. The code for that is not complicated, but I need to add a comment to explain it (because it is not obvious at first sight what is happening).

    let kmer = &sequence[i..i + ksize];
    let krc = &rc[len - ksize - i..len - i];

Funny bits:

• It is a gigantic PR because the ComputeParameters setters and getters are... a lot. But it's also very repetitive code (and maybe could become a macro, but then it becomes harder to reason about/fix bugs later).
• [RFC] making compute CLI command usable inside Python #245 explodes the args in the compute function signature, but since Rust doesn't have default arguments for functions I went in the other direction there (collapsing all the args into a ComputeParameters struct). Python's default arguments are so nice.

Previous blurb:

I'm using this branch for implementing decoct compute and figuring out how to:

1. expose functionality in a way that doesn't requiring moving all the Python compute code into Rust
2. allows some parallelization if using the Rust compute parts (implemented with rayon)

Checklist

• Is it mergeable?
• make test Did it pass the tests?
• make coverage Is the new code covered?
• Did it change the command-line interface? Only additions are allowed
  without a major version increment. Changing file formats also requires a
  major version number increment.
• Was a spellchecker run on the source code and documentation after
  changes were made?

[codecov]

Codecov Report

Merging #845 into master will decrease coverage by 12.29%.
The diff coverage is 64.73%.

@@            Coverage Diff             @@
##           master     #845      +/-   ##
==========================================
- Coverage   91.19%   78.89%   -12.3%     
==========================================
  Files          69       82      +13     
  Lines        4916     6993    +2077     
  Branches        0      479     +479     
==========================================
+ Hits         4483     5517    +1034     
- Misses        433     1174     +741     
- Partials        0      302     +302

Flag	Coverage Δ
#pytests	90.54% <98.48%> (?)
#rusttests	47.29% <15.21%> (?)

Impacted Files	Coverage Δ
src/core/src/signature.rs	33.73% <0%> (ø)
sourmash/_minhash.py	97.21% <100%> (-0.13%)	⬇️
sourmash/cli/compute.py	100% <100%> (ø)	⬆️
sourmash/exceptions.py	79.41% <100%> (+1.28%)	⬆️
sourmash/signature.py	87.67% <100%> (+1.07%)	⬆️
src/core/src/sketch/minhash.rs	32.2% <18.91%> (ø)
sourmash/command_compute.py	97.2% <99.1%> (+1.04%)	⬆️
src/core/src/from.rs	71.66% <0%> (ø)
src/core/src/lib.rs	51.63% <0%> (ø)
... and 18 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d25d786...69ce7a6. Read the comment docs.

[ctb]

I held off on fixing sourmash compute to use FileOutput (#853) because of this PR, but you might want to consider tossing it in here. Or I can do a PR into this branch. lmk.

[luizirber]

I held off on fixing sourmash compute to use FileOutput (#853) because of this PR, but you might want to consider tossing it in here. Or I can do a PR into this branch. lmk.

I updated it in 03bd393, does it match what you did in the other commands?

Also ready for review and merge @ctb

[ctb]

nice work!

[ctb]

maybe change output_fp to something like output_filename here?