Extended dissipativity state estimation for generalized neural networks with time-varying delay via delay-product-type functionals and integral inequality