[PATCH] Model Deallocator node in CFG

[Jordan Rose]

  I actually think we can treat the deallocator as part of the 'delete' expression itself, rather than modifying the CFG. It may be a bit asymmetric, but the only real reason to have the separate allocator element in the CFG is so that it gets properly sequenced with the constructor. For the delete expression, the natural place to evaluate it is after the destructor is called, which already has a separate CFG node. What do you think?

http://llvm-reviews.chandlerc.com/D2647