[PATCH] [StaticAnalyzer]Handle Destructor call generated by C++ delete expr

[Jordan Rose]

  I'm seeing crashes in test/Analysis/new.cpp due to unexpected ProgramPoints in generated node for the null case. I'm not sure we expect the previous node to be propagated forward; notice in the initializer case there's a comment for "Construct PostInitializer nodes whether the state changed or not, so that the diagnostics don't get confused". I haven't looked into what the Pred node is, though. This needs to pass the regression tests before it goes in!

http://llvm-reviews.chandlerc.com/D1594