Implement numeric fallback

[emberian]

Doesn't yet converge on a fixed point, but generally works. A better algorithm
will come with the implementation of default type parameter fallback.

If inference fails to determine an exact integral or floating point type, it
will set the type to i32 or f64, respectively.

Closes #16968

[rust-highfive]

r? @brson

(rust_highfive has picked a reviewer for you, use r? to override)

[flaper87]

shouldn't this be changed to something that actually tests multidispatch & ambiguous destination ?

[emberian]

Maybe, I don't really know what that means or what the test would be though :)

[tbu-]

CC me

[emberian]

@nikomatsakis there are now failures:

/home/cmr/rust/src/test/run-pass/traits-multidispatch-infer-convert-source-and-target.rs:35:5: 35:9 error: unable to infer enough type information about `_`; type annotations required
/home/cmr/rust/src/test/run-pass/traits-multidispatch-infer-convert-source-and-target.rs:35     test(Default::default(), Default::default(),  2, 4);
                                                                                                ^~~~
error: aborting due to previous error
/home/cmr/rust/src/test/run-pass/traits-multidispatch-infer-convert-target.rs:42:5: 42:9 error: unable to infer enough type information about `_`; type annotations required
/home/cmr/rust/src/test/run-pass/traits-multidispatch-infer-convert-target.rs:42     test(Default::default(), 44_u32, 2, 4);
                                                                                     ^~~~
error: aborting due to previous error

[alexcrichton]

If ok, it'd be nice to not jump the queue for backwards compatible changes like this. For example I've been trying to land the prelude stabilization for over a week now but it conflicts with just about everything. With only a week left until the alpha, we need to prioritize backwards-incompatible changes first. Thanks!

[emberian]

Yes, sure. I've been manually scheduling PRs that had a priority above this one, didn't notice that that one was huge.

[steveklabnik]

(on the plus side, I've been waiting for this to land forever, because it impacts docs very heavily, so I appreciate the jump)