Also:
Most of the competitive peeps around here who pull in flexion pull in braced flexion. They're in control of it.
This situation ^ is quite a bit different than that of the random dude who starts with a near neutral spine (or with moderate flexion) and flexes further to end-of-range t-spine flexion during his pull (dangling off ligaments).
The first scenario is probably perfectly safe; the second scenario turns my stomach a bit
Since you'd have to extend at the top to stand erect, wouldn't you actually get a concentric component for both of these with a roundback deadlift? Which to me isn't less of a stimulus, just a completely different one. And one that is useless for the purposes of a novice training program.
If anything, I'd actually think the hip extensors themselves are shortchanged, as the big benefit to a roundback is hips closer to the bar at the beginning of the lift. You can usually pull more weight, but whereas a completely straight back pull can be hard off the ground, most people find the roundback is fast off the floor but then stalls, and is likely to fail, closer to lockout. That implies that with roundback, you will never be able to achieve a weight that stresses the hips at the same intensity as the flat back pull, as the upper back gives out first. And then you aren't training your back in a pattern more relevant to most activities.
Quibbling aside... I do think the moral of the story is that roundback pulling is something that should be saved only for those who have the experience to perform it correctly, and the drive to compete enough strength competition such that they need to find their strongest pulling stance. I think for anyone else, flat back is more useful for the job we seem to want the deadlift to do.