Polynomial approximation of distributed lags and linear restrictions: A Bayesian approach