Maximum State Entropy Exploration using Predecessor and Successor Representations