Joining the debate after looking at the videos...IMHO,
In the McNish accident, the gravel trap didn't "trap" the car (it would have probably sent it rolling violently if it had), but equally it didn't get properly airborne, as it skidded/skipped across the gravel leaving a pretty constant plume of dust behind it. The tyre wall, armco and the structure of the car did their job by absorbing a huge amount of kinetic energy quickly, but in a relatively controlled way; the loose wheel had little energy left when the photographer was running away from it. That it went into the barriers at the angle and height it did, and that nobody got hurt may well have involved a certain element of luck.
In the Rockenfeller accident, the quick spin left into the barrier was probably more down to the contact from the Ferrari moving right against the middle/rear left of the Audi, which put both right wheels well onto the grass; combined with bags of grip on the left side wheels on the tarmac, the dynamics are inevitably going to slew you left - very quickly. Again though the car didn't get properly airborne.
In both of those cases, the dynamics of the accidents combined with the structure of the car left the drivers unscathed. I have yet to hear mention of the third serious accident of last weekend at 0330hrs, where the video shows Mike Wainwright driving the No 60 Aston Martin Vantage running wide onto the grass outside the right hander after the first of the Porsche curves, slewed left in front of another car (without making contact with it), hit the opposite side barrier initially with the front right of the car which spun the right side of it into the barrier. Which was a solid concrete wall. Which resulted in (reported) injuries of broken ribs, a punctured lung and a fractured pelvis.
A lesson or two there perhaps about LMP v GTE structural designs, probably something about fins, but more particularly about the relative energy absorption properties of tyres/armco, armco alone, and concrete.........
MG Mark