Code for our OTML NeurIPS paper : 'On Combining Expert Demonstrations in Imitation Learning via Optimal Transport