Would the published figures suffice to choose between methods i) and ii)?
Prior to recrystallization, there is 0.94 of the major enantiomer and 0.06 of the minor.
After, 0.91 (0.87/0.96) of the previous step remain and contain 0.90 of the major and 0.01 of the minor. So 0.05 of the minor was removed while losing only 0.04 of the major.
If an accurate 50/50 recrystallization happens despite the solution is strongly mismatched, then the figures are too close to call. But if the solution mismatch spoils the 50/50, then the recrystallization must operate instead by adding preferentially the major form on itself.