Accelerated Distributional Temporal Difference Learning with Linear Function Approximation