updating tolerance on angle_deriv function from 1.0e-3 to 1.0e-8. This was causing major issues in derivs for a new Mg energy function I was developing, and took a while to track down!