RSSA backend

[Stephen Weeks]

> One note, that I mentioned before, is that we're getting
> -d 1 labels corresponding to RSSA rather than corresponding to SSA.
> I think this is bad, both because -p saves .ssa and not .backend.machine;
> furthermore, even saving .backend.machine with -keep-pass backend isn't
> quite right, because backend.fun introduces yet more Machine.Block.t's, so
> the labels in the profiler are sort of in limbo.
>
> Thoughts? 

One possibility is to do admit that profiling is wrt RSSA, not SSA.
That is,

1. add -keep rssa, which would save the rssa after insertArrayInits
2. have -p imply -keep rssa, not -keep ssa
3. have -keep dot cause dot files to be saved associated with the rssa

That fixes the all the problems, doesn't it?  The drawback is that
RSSA is harder to understand than SSA.  I don't understand your point
about using -keep-pass backend being wrong -- .backend.machine seems
to exactly correspond exactly to the profile labels.

Anyways, I am also happy with pushing back the profileInfo into RSSA.
It can't be that hard.

> Any reason not to mention the -color option in the profiling
> documentation? 

Nope.  Go for it.  Also, reorder the -p entry in the man page so it is
before -polyvariance.