I don’t have the turnout numbers for 2006 on hand. And unfortunately solving it in that direction is not as simple (you have two equations with three unknowns).
43D, 38R, 19I in 2006. But this was a mid-year election, where many PA voters who voted in 2004 did not vote in 2006. The 2004 breakdown, which also happens to be how the state actually voted in both 2004 and 2000 (the last two previous Presidential electoins), is far more relevant to projecting turnout by party ID in 2008.