There are arguments both ways. I actually think that situation has to dictate that move. I agree that Davey was going to hesitate pulling LaRoche pending extra innings. If you are still one run down, he is likely to pinch run, because you have to score one. Once the game was tied up with no outs and guys on 2nd and 3rd, there's no pressing need to pull LaRoche. The thinking was likely that of all the things that could happen to keep a run from scoring, the margin between the fast guy and LaRoche isn't big enough going 90 feet. Anything that would bring in Kobernus from 3B would likely bring in LaRoche. If he's on 2B the margin is bigger. Things that would go wrong would be a series of Ks, pop ups, an easy DP at home. All those would strand the fast guy too. The only scenario is that only one sac fly opportunity occurs, and it is at the precise distance to nail LaRoche but not the faster runner. That's a relatively narrow window. So, the manager has to weigh that versus the possibility that there will be successive failures that score nothing, and now the defensive glove and hot lefty bat is gone.
With Desmond behind LaRoche, an event that gets an out at home plate leaves another fast guy on 3B with the subsequent attempt. Which actually brings up another argument that the mistake was really not sending LaRoche initially on Demond's double to force the issue. You either win then, or you leave Desmond on 3B with two more chances to bring him home. But from my vantage point, I figured a good throw has LaRoche dead to rights, so I liked him being held with no outs.
I would have PR for him at any base if they were still losing. But I don't think there was a pressing need to do it after the tie game unless the strategy was to try and go small ball. In hind sight, PR would have scored from1B on the double. But you don't know that's going to happen, so he opted to wait to preserve the player and not burn a guy off the bench heading to extras. Once he got to 3B with no outs, there wasn't as much risk.