I think this will happen naturally as long as we pay attention to the Python tests; there are a lot of tests for corner cases at fairly high API levels (usually the CLI). An unanticipated benefit of functional tests over unit tests I guess? :)

I do think this provides a pretty good argument and incentive for more strongly prioritizing integration of branchwater-plugin-style speed and functionality into the core sourmash package. FBFW, the sourmash Python layer is pretty thoroughly tested, while the branchwater plugin is pretty light on tests.

from luiz: rust and python should support same ops. I started unifying a bunch of the code in #2728 but at this point that's very outdated

Note this was a comment on KmerMinHash/KmerMinHashBTree on the Rust side, not across Python/Rust.

But it is connected on the Frozen/Mutable discussion between Python/Rust, since in general Frozen == KmerMinHash and Mutable == KmerMinHashBTree