Given the 'lenient' limits granted the Autopilot in cruise, and bunk weather, it is reasonable to assume the pilots were established in "acceptable" autoflight until sometime just before the stream of ACARS signalled what is likely an already upset aircraft. The mechanical (A/P) limit having been breached, the pilots were left with a rapidly manouvering airframe with which they had not had time to acommodate their 'meat-think' (
). As per ZeeDoktor, the transition between a docile machine and one that requires a stable set of anticipatory thinking only, the conclusion is quite apparent. (That this is a flaw in AB philosophy)
If impact was 6.5nm away from last known position, it would need a bit more forward travel past LKP to avoid considering the a/c dropped like a stone at an average AOD of 45 degrees. This means a turn (back) of some description, and most likely pilot input. This excuses to some extent the searches assumption of distance away from LKP.
A "flat impact" is concluded, but this is not to say that she wasn't in some Pitch extremis just prior to impact. If a tail-first hit, the VS driving downward in its aft spar axis explains the lower corner (Rudder) damage, and could allow for the reasoning that concluded the VS landed away from the crash site.
This also allows for some 'time' for the Mains to sever their stowed position downward, just prior to contacting the sea essentially vertically, and perhaps without a horizontal vector to speak of.