This illustration is what the Air Force Museum is using to depict the "idiot loop" that the Air Force developed in 1952 to more or less safely deliver a nuclear weapon from a tactical aircraft. Safe in terms of both ingress and egress. The problem is, it's depicted as a loop and the actual maneuver is a half-Cuban eight. It starts as a loop (a loop is what you'd be doing on a Ferris wheel if the car you were riding in wasn't free to dangle) but after you go over the top, inverted, you do a half roll back to upright flight while diving back to treetop altitude to depart the premises as fast as possible back along the route you took to get to the target. A loop, as depicted, would keep you in the vicinity of the upcoming blast for too long...
After I wrote this, I discovered that there were proponents (at least one in the Navy, anyway) for the loop recovery from the over-the-shoulder mode and it was evaluated by him at China Lake in an F7U-3. There is a video of it somewhere on the interweb which I can't now find.
It was compatible with LABS since the pilot was free to recover from the inverted climb that he was in by any maneuver he wanted to use. However, in addition to probably not providing as much separation as the half-Cuban eight (I don't have any numbers to compare the two), completing the loop before encountering the ground was not entirely risk free, whereas the half-roll in a dive was. Remember that having enough altitude coming over the top to complete the loop was compromised by having to lug a heavy, draggy bomb a good part of the way up and a dive to build up loop entry airspeed was precluded by both the LABS computation and the need to make ingress at treetop level. The F7U-3 was probably better off from this standpoint because its engines were equipped with afterburners.
In any event, I haven't found any operational descriptions of the use of a loop on the interweb or in the literature other than this report.
This is an illustration of the three LABS alternatives from my book, Strike From the Sea.