[PATCH] Fix to PR15537 - assertion failure when comparing atomic unsigned int to int.

[Richard Trieu]

  I'm not the right person to review this as I've only done cleanup work in this area.  That said, I do have 2 questions:

  1) Does this affect any other operations involving atomics or just comparison operators?
  2) This seems like something UsualArithmeticConversions should handle.  Would it be better to fold it into there?

http://llvm-reviews.chandlerc.com/D599