I would assume that since the source material was 24 fps, transferring to Super8 would have to be 24 fps. Otherwise it would be jerky as hell. I think many (all?) Super8 cameras might have been 18 fps, though.

As for editing and making MPEG-2, I'd be very careful to find out beforehand exactly what Vegas does in order to make DVD-spec video. Does it internally render 29.97 fps, and encode as such, or will it pass true 23.976 fps and let TMPEGEnc add flags?

I'm more a video than audio guy, so my vote would be to subtly tweak the audio to match the frame-accurate video. I think you open a whole new can of worms by trying to sync audio and video by duplicating film frames. In my opinion, you'd be better off to take frame-length chunks of audio out here and there, to match the video.

Also, keep in mind that 24 fps is not 23.976 fps is (probably) not 23.98 fps -- those differences right there may be enough to make the audio drift!