You may be suggesting that the load carrying surfaces of the rails are slightly canted towards each other. I suppose that if toe in is not possible, as in the case of a train, the next best thing would be to simulate the effect in the rails. Am I on the right track? Do I need to switch my thinking at some point?

It had occurred to me that because the engines always turn at the same speed no matter whether the train moves or not, that this would prevent torque from tipping the train over when it starts up, but that's too easy.