[CUDA] Conservatively mark inline asm as convergent.

Summary:
This is particularly important because a some convergent CUDA intrinsics
(e.g.  __shfl_down) are implemented in terms of inline asm.

Reviewers: tra

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D20836

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@271336 91177308-0d34-0410-b5e6-96231b3b80d8