[PATCH] [SCCP] Calculate Null Pointer Comparisons

[Nicholas White]

> Please provide a real description of the change.
I've updated the review description - hope it's OK now!

> You can replace dominated uses
Ah yes, I've carried that functionality forward from the SCCP version
of the patch. Unfortunately the SimplifyCFG pass has been updated to
use the new pass manager, but the post-dominator tree analysis hasn't,
so this patch now converts it to too (so it's now +250 -157). I've
also added another unit test to prove that it doesn't do the
optimisation when the branch doesn't post-dominate the test.

Thanks -

Nick