I think TA was saying that if we look at this graph, we have to conclude that there's been a semipermanent structural change in the economy (where the previous level of production can be maintained with a higher unemployment level) or GDP is actually a crappy gauge of economic health.
I'd point out that GDP is still way below trend (where it would be if the upward slope had kept going), which means there should still be a lot of unused resources out there, including people without jobs. Also, in the two most recent recessions job recovery lagged behind GDP recovery, which it hadn't done before. So it's probably not the "new normal" in that 91% is now maximum employment in a growing economy, but it's gonna be a long slog back.
Brad DeLong posted about jobless recoveries on his blog during the early days of the crisis, but I haven't been keeping up with him enough to effectively steal his opinions and pass them on as my own. Last I checked, though, he thought he had a good handle on why employment bounced back quickly after the other post-WW2 recessions, but no real answer as to why that had changed.