Replacing C++ with Rust

[luizirber]

This PR replaces the C++ parts of sourmash with Rust.

Tests are passing, but some optimization might be good (especially on avoiding too many calls/memory copies across the FFI layer).

Other branches/PRs depending on this one:

• Moving loading and save sigs to Rust #532 starts moving more of the Python implementation to Rust: signature loading and saving. This removes the dependency on ijson, but I decided to keep it separated from this PR because the Python signature loading/saving code is fine (but I needed it implemented in the Rust side for the webassembly parts).
• There is a nodegraph implementation in Rust, exposed to Python in the rust_nodegraph branch. Still need to fix and test more before turning into the default, but would also mean optional dependency on khmer.

There are more details in https://github.com/luizirber/2018-python-rust (the poster presented at BOSC and SciPy 2018).

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-io]

Codecov Report

Merging #424 into master will increase coverage by 1.07%.
The diff coverage is 89.58%.

@@            Coverage Diff             @@
##           master     #424      +/-   ##
==========================================
+ Coverage   78.52%   79.59%   +1.07%     
==========================================
  Files          43       45       +2     
  Lines        6534     6588      +54     
  Branches      499      454      -45     
==========================================
+ Hits         5131     5244     +113     
+ Misses       1100     1043      -57     
+ Partials      303      301       -2

Impacted Files	Coverage Δ
src/index/bigsi.rs	60.86% <100%> (+1.31%)	⬆️
sourmash/sig/__main__.py	94.45% <100%> (-0.41%)	⬇️
sourmash/sourmash_args.py	96.73% <100%> (+0.01%)	⬆️
sourmash/sbtmh.py	85% <100%> (+0.1%)	⬆️
sourmash/__init__.py	100% <100%> (ø)	⬆️
sourmash/_compat.py	100% <100%> (ø)
src/index/sbt/mhbt.rs	41.37% <40%> (-0.16%)	⬇️
src/sketch/minhash.rs	37.92% <41.66%> (+0.2%)	⬆️
src/signature.rs	37.58% <66.66%> (+1.52%)	⬆️
sourmash/utils.py	71.92% <71.92%> (ø)
... and 7 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 93b9b90...320b021. Read the comment docs.

[luizirber]

Turns out there is only one place that uses the common set returned by intersection: the test method =P
ccc05ec#diff-acb3fd27a7c58bb1a046d1a231f2d6cdR277

While we don't have anything depending on this behavior I suggest we add an in_common argument (defaulting to False) to avoid having to build and output the common set.
ccc05ec#diff-2032759ae736988ba55922787626efb5R281
(I still need to do the rust wrapping around the common set, that's why it is a weird mixed implementation for now)

[ctb]

sounds good - issue or PR or ??

[luizirber]

@ctb: covered the intersection changes in #453

[luizirber]

Ready for review and merge @ctb

[olgabot]

OMG WHOO!!!! Congrats!!!