Lint: Multiple inherent impls

[llogiq]

We may want to have a lint that checks for multiple inherent impls. Why those are not problematic from a language standpoint, they can split the implementation of a type and make the code harder to navigate.

The hardest part is that the lint needs some storage, namely a set of type names that already had inherent impls, but we already have examples of that.

[oli-obk]

We do have this lint now for non-generic impl blocks. The next step would be to make the lint trigger on any two impls blocks with the same generics.

[clarfonthey]

I do agree that this should apply if the impls are in the same file; it makes the most sense to me to replace impl bounds with where clauses on methods. That said, it also makes sense to be able to do multiple impls in multiple files— one example I've used is that the nibble crate has a separate module for implementing formatting traits and passing through inherent methods. The actual types are spread into multiple modules for readability but these impls are similar enough that it makes the most sense to use a macro.