> Can we learn from the diversity of those implementations

We can learn that folks will reimplement the same thing for a variety of reasons - some to learn, some to mesh with some other system, some to have control/knowledge of the solution, because the existing ones look unsupported, or out of ignorance of other implementations.
And they will publish them to crates.io regardless of whether the ecosystem benefits from it, because the rust book told them to, it's ever-so-slightly easier to use crates than git repos as dependencies of other crates, and so they can put their crate count on a resume.

> and come away with some best practices?

there may be some good implementations in there, but most will just be different for no reason you'll be able to extract any useful insight from