[tree-ssa] Loop analysis

[David Edelsohn]

>>>>> Jason Merrill writes:

Jason> I assumed that the current position of the loop calls just had to do with
Jason> only needing a single #if 0 block.  Presumably when we actually run the
Jason> loop optimizer we will think about when it should run.  Complaining about
Jason> the position of disabled code seems a bit premature.

	The original patch did not have the code disabled and inserted the
code in the exact same location.  Diego later asked for it to be #if 0'ed
because no loop optimizations were implemented.

	While working on SSA loop optimizations, I need to know if the
interaction between the existing loop optimization infrastructure and the
Tree-SSA design is hobbled from the beginning.

David