We know that a lower V1 gives a higher accelerate-go distance. (Contrary to what I first wrote in this thread - oops!)
This program puts out min and max V1's, since apparently a range is allowed. Can you explain what "policy V1" is? It seems to me that it's the actual single V1 chosen (from the range) to use for this particular take off.
Well, it would naturally make sense to me that the accelerate-go and accelerate-stop distances would be based on the single V1 chosen. (In that case, the policy V1's are almost the same, which should yield a longer accelerate-go that we expect to see for the reduced thrust, but we don't actually see.)
But instead, what seems to be happening is that the distances are based off the worst-case V1's for each thrust situation, and not the policy V1's. This allows a different V1 to be chosen from the range without having to re-run the calculation. To me, this explains it: for the reduced thrust, min and max are both 140 so there's no range. But the full thrust the min (I.e., worst case for distance) is waay down there at 94, thus jacking the distance up.