State estimation for delayed neural networks with reaction-diffusion terms: Hardy-Poincare inequality