Reproduction: Revisiting Zeroth-Order Hessian Approximation: A Single-Step Policy Optimization Lens