I'd love to be able to find it, but I've looked in the past and come up a blank, but they did a study with it on a bowling machine firing balls down an empty net and there's not much to suggest that it's significantly wrong (if anything I think the results suggested the margin of error with umpires call is too large).
I'd be inclined to say that the issue is more likely an optical illusion caused by foreshortening and the camera angle on TV broadcasts than a widespread failure by Hawkeye.
In this instance, let's not forget the umpire gave it out on the pitch and England went up for it loudly, too, so I don't think it's particularly indicative of a failing of the technology that it showed it smashing in to the stumps.
I seem to remember the powers that be mentioned the onward path of turn was inconsistently projecting. Think I heard that earlier this Summer actually.
My initial reaction to this appeal, as many of that length example have been, was too high (As well as impact in this case). The trajectory of a quick hitting the pad so far forward from that pitching length would take the ball nearer the top of the stumps.
I'd like to see how the batting crease & the batsman position at the time of impact is taken in to account. In theory it doesn't matter where the batsman is when the pad is struck, it's where the ball pitches length wise & the judged trajectory from that length compared to an average trajectory for that same length if required for fuller impacts.
Just a feeling I've had for a while that front foot contact from a good/shorter than good length from the quicks aren't projecting a high enough bounce. Perhaps it's an illusion as you say.