See my post above. Your tidbit from Gallup was a great find. I just fleshed it out.
The suspense is building...I think you are right. Either the 2004/2010 turnout model could swing the polls 5+ percent from one side to the other.