implement batching algorithm in Rust

[dave-doty]

Cython appears to have changed significantly since we first implemented this, and many people (including me) are having trouble installing ppsim.

Rust is likely to be faster than Cython, and given how good that community is at making reliable tools, I think PyO3 is likely to be more straightforward to use to link the fast pre-compiled Rust code to the Python library.