RFR: Eliminate LoadAddr edge from LoadReferenceBarrier

[Roland Westrelin]

> I wonder if it makes sense to pursue that and the safepointing issue
> (which seems important), or go for late barrier insertion to begin with,
> which would solve both. I have no idea about the complexities of either
> path, and also if late insertion has any new problems of its own. I'll
> leave that to you ;-)

I'm unconvinced it's worthwhile to pursue late insertion. It seems
solving the safepointing issue is simple enough. So I'm not sure what
late insertion really buys us while it's quite possible it won't be
straightforward.

Roland.