Cross-crate overloading doesn't work

[catamorphism]

as per a FIXME in core::num

[eholk]

Do we know this is the case? I'm pretty sure we are using this now.

[catamorphism]

I assumed this was true because core::num is pretty new, but I'm not sure. Is there a test case?

[eholk]

I've got a bunch of code that implements + for [T] as an iface. I haven't noticed any cross-crate problems with it. I think we had some problems along these lines when I started this summer though.

[catamorphism]

Ok, well, I still want there to be at least one test case (possibly an existing one) before I close this!

[pcwalton]

I wrote that because @nikomatsakis had said he had found a bunch of issues with it, mainly in reachability checking IIRC (although I also remember he had trouble reproducing them). Paging him.

[nikomatsakis]

These problems may have been fixed in the meantime. They were quite easy to reproduce: remove the get_elt() method from dvec and start using [].

[pcwalton]

Added a (working) test in 6a7c714; closing.