Well, the cement must not have stayed clear for very long - the PAL 35mm telecine bootleg of Empire that VideoCollector obtained a few years back had the same kind of yellowish cement gunk on at least one splice. And that was from a first-run UK theatrical print. (-1 made mention of having to crop out glue from the grind house transfer, I assume their prints are also first-run?)

I guess it was just something not meant to be noticed and usually cropped out in projection.

Also, I have read that A roll/B roll were printed from the negative by printing the A roll (with the slugs in between coming out as black), then printing the B roll onto the same film, exposing the B-roll shots on the black sections between the A roll shots. I can't find anything about interpositives being made in pieces and glued together.

If the negative was A/B, that would mean the yellow blobs are printed in, but that doesn't make sense. For one thing, it would mean the cement was actually light bluish-gray, which seems odd. More importantly, it wouldn't explain why different sources made at different times have different marks. They wouldn't be ungluing and regluing the negative multiple times, if the splices are printed in from negative then only the shots which were replacements would have different splice marks. The more I think about it, the more confused I get...