[tree-ssa] Lazy updating of stmt operands

[law@redhat.com]

In message <20031207174622.GA13528@atrey.karlin.mff.cuni.cz>, Zdenek Dvorak wri
tes:
 >1) Merge_blocks.  The natural way (for virtual operands, the only way) of
 >   getting rid of the degenerated phi nodes in the second block is to copy
 >   propagate the set implied by the phi node.  Since cfg cleanup should
 >   be fast, I don't like scanning of all statements as implied by
 >   compute_immediate_uses.
Before spending a lot of time worrying about this, I would first look into
why the combination of the dominator optimizer and DCE doesn't kill such
PHI nodes.

While I'd like to have some generic code to optimize away useless PHIs, in
practice I've haven't found the need.

jeff