Something about this doesn't add up; i think you have misassigned the products.
Think about the initial example, 1:1 ratio of epoxides (regardless of NMR assignment). Due to the slow reaction rate, and poor selectivity, this is assigned as a case of mismatched stereoinduction between the catalyst and allylic methyl.
So, changing from the (R,R)-tartrate to the (S,S)-tartrate gives a vast increase in rate and selectivity. Which you expect if you now had matched stereoinduction, with both the catalyst and the allylic methyl group giving the same sense of stereoinduction. Which would occur if the major product was the undesired isomer. So i think you probably have your assignments backward.
You could investigate this by running the reaction with (R,R)-DIPT, to see if you get another mismatch ie. slow reactivity and poor selectivity. If you do, you assignments are probably backward. Alternatively you could try to induce crystallisation of the separated major isomer. You'll struggle to get crystals with that TBS group, my instinct would be to knock the TBS group off and try to introduce a crystallising group (tosyl or m,m-dinitrobenzoate ester are a good start). Truck your crystals off to the nearest friendly crystallographer and hopefully they can tell you for sure.
PS. a stupid question, but all the best ones are - you are certain you have the (E)-alkene?