Wrong posterior in multidimensional GaussianRandomWalk

@jessegrabowski, got it. It makes sense. Thank you for your assistance.