I'm not saying this is a practical case in the real world, but part 25 doesn't explicitly prohibit rotation starting before the in-air stall speed.
If we assume that the in-ground effect stall speed is Vmu (I know they are not) then in order to comply with part 25 I need to
(a) be able to get to V2 (min 1.13Vsr) before 35ft
(b) have a VLOF of not less than 110% Vmu
So I need a situation where I start rotating (well) before Vsr/Vmu BUT don't get airborne until 110% Vmu and don't get to 35ft before 1.13Vsr.
One thing that could do that is a mis-match between the ability of the aircraft to accelerate and the ability to pitch.
Suppose for some ungodly reason you decided to do a takeoff using only the stab, not the elevators. The stab rate is insanely low compared to the elevators, so you need to start trimming nose-up early - there's your early rotation. It's still going to take forever to get to VLOF, so we'll easily meet the 110% Vmu rule, and our "takeoff" will likely be at very low angles, so we'll take a while to get to 35ft, by which time we'll be well above min V2.
Now, that's not a terribly practical aircraft, but it would (at least in regards to Vr) be compliant to the letter of 25.107.
Could there be some kind of high T/W limited pitch authority aircraft design - maybe one of these supersonic ideas floating around??