[tree-ssa] Out of SSA status and issues

[Richard Kenner]

    Suppose there is register pressure and we are in the register allocator
    and decide that p1 doesn't get a hardreg.  If you now don't allow the use
    of p1 to be replaced with *p2, you are forced to emit such code:

Of course, but we weren't talking about the register allocator.  Instead,
we were talking about a high-level optimization.

The register allocator indeed needs to know what a psuedo corresponds to
so that it can undo the optimization is there aren't enough hard registers.
We do some of this, but not enough.