Convergence to Nash in the potential Linear Quadratic games and accelerated learning in games