I wonder if the statements about a slow take-off roll can give any clues to the "wrong take-off configuration" hypothesis.
IF the flaps and slats were retracted, one would expect the acceleration to look normal (possibly even slightly better) until the airplane is rotated (at a speed below Vr for the no-flaps configuration). From there on the acceleration would suffer due to the nose-up angle and ensuing high drag, prolonging the long ground roll before finally becoming airborne (as it briefly did).
IF on the other hand the take-off configuration was okay and the airplane still had a longer than normal take-off roll, then it would point to a thrust deficit below commanded thrust, and the abnormally slow take-off roll would be apparent already before it rotated.
The question is then, did the witnesses comment on a slow take-off in the early stages of it, or only after rotation? Anyone seen such statements?
A sub-question for the second case: What faults would possibly cause a thrust decrease and still not show up on the engine instruments or alarms?