Near-Optimal Policies for Dynamic Multinomial Logit Assortment Selection Models