This doesn't fix all of the clippy warnings in these crates. There are
decisions that needs to be made in there that I'm not willing to make.
Signed-off-by: pep <pep@bouah.net>
That way, callers can put multiple candidate implementations in, for
example, a sorted vector and more efficiently select candidates to try
when looking at a new element.
By removing the implicit `Sized` bound, they can be used with `dyn _`.
The `Sized` bound is unnecessary for a boxed value, so it is a
superfluous restriction. Also, the use with `dyn _` may turn out useful
in the future when we want to allow type-erased XSOs.
The main lib.rs is getting a bit cluttered, so I'm trying to bring some
order into the chaos by moving some things into other modules.
skip-changelog, because there are no user-facing changes (names which
have been moved are doc(inline)'d and pub use'd in the main lib, and
other things (trait implementations) aren't addressable by users).
This declutters the main `xso` namespace. In addition, if (when) we
introduce more complex generic implementations, we might want to have
tests for these, and those can then live there, too, without making the
main `lib.rs` file gigantic (or moving the tests too far away from the
tested code).
