RFC: Define binary operator traits using associated fns, not methods.

[nikomatsakis]

There has been a constant stream of confusion relating to the interaction between auto-deref and various overloaded operators, particularly Eq (e.g., #6257). I think we ought to define the binary operator traits using an associated fn, not a method. This would side-step any issues around autoderef, and also helps to make the method resolution clearer.

For example, Eq would be define as follows:

trait Eq {
    fn eq(a: &Self, b: &Self);
}

meaning that x==y would be expanded to:

Eq::eq(&x, &y)

The same would apply to other binary operator traits.

[nikomatsakis]

nominating for backwards compat milestone

[nikomatsakis]

I think I withdraw this idea. We can just add functions if we think that'll be useful for macro authors.