Not sure, but the 2:30 overlap could be if the first recording goes to 2:30:59 and the second recording starts at 2:30:00.. That 59 seconds could account for the overlap.
Again, not sure at all, but it's a possibility.. As far as the original issue.. Haven't got a clue, but it sounds like the scheduler didnt release the tuner. Probably a software bug..
If you do the double RBR, maybe do a test run with random recordings and see if you can replicate it... Curious if it happens again.