Re, my exploration of CONCOCT: I fear this one is a harder sell in terms of rewriting it in Rust. Contrary to initial assumption, the heavy expensive stuff is written in C, not Python. It's proving quite resistant to optimization. It can be made a bit faster if I threw SIMD and such in there, but C can do that too. Authors knew what they were doing. It might genuinely be better to just turn to my "phase 0 baseline" and just consider that an improvement in terms of packaging. I'd be happy to explore a Rust rewrite more, but we'll need to nail down clearly what exactly Rust's added value is in this case - it'll be more on the side of correctness and maintainability than performance. Otherwise... let's just see if I can improve packaging, or just leave it at that.