[libc++] Check if we can use ifunc to resolve operator new[] & friends at link-time

Right now, `operator new[]` & friends often turn around and call `operator new`. If we could resolve at link-time whether the operators have been overridden, we could potentially use resolver functions to have `operator new[]` resolve to `operator new` directly, removing an indirection / jump.

Since this is really hot code, it might be worth it. Idea by @philnik777 .