cps & contification

[Matthew Fluet]

> > I also added a proof of the claim that
> > A'(f) = Uncalled iff there does not exist a path from fm to f.
> 
> Here is a shorter proof of half of the claim

Nice.

> Matthew, are you interested in doing the analysis implementation?  It shouldn't
> be too bad given that the dominators algorithm is already there.  If you do, it
> might be nice for you to catch up with the latest snapshot, which contains a new
> CPS IL, in which many of the lists have been changed to vectors.

Sure.  I'll try to set it up the way I attempted to earlier, with
diagnostics for the three analyses.  That should be useful for comparisons
and benchmarks.