Thanks everyone for the input. I managed to remove the old seals without damaging the seals or the aluminum nuts, then pressed them back together using one of the old seals as a spacer, as Bryan suggested. This worked great.
I decided to pull the spacer and check the O-rings. Part of my motivation for pulling the transaxle was the fact that it was leaking like a sieve. It would drain itself down to the output shaft seals in a few hours. Upon pulling the finned nuts and looking at the seals, there wasn't anything about the seals that seemed to be an obvious cause of the leak, so I figured this was reason enough to check the O-rings under the spacers. One of them seemed to be seated a little bit outside of the slot and had some wear from riding on the splines, so I decided to remove both.
When I removed the nuts a couple months ago I took a video checking how many turns each one took to remove, that way I could avoid changing the bearing preload and crown wheel backlash as long as I left the case together. After pulling the spacers and installing the seals the right direction I found that there is no room to adjust the crown wheel position left to right in the case. Instead, the nuts both go back in exactly the same number of turns that it took to remove them (about 8 and 1/2 turns each), and then it feels like they're fastened against a completely rigid surface. Loosening one doesn't give any more room for the other one to go further in.
I did measure the number of turns to get the right bearing preload before doing the final case assembly (although at the moment I'm unable to find my note where I wrote it down). I reused the original bearing, so setting this was pretty trivial. Backlash feels pretty low with everything assembled, but I'll borrow a dial indicator from work to check if it matches the 0.005-0.010" spec in the manual. Any ideas for why there's no room to adjust the backlash?