Exploiting Causal Graph Priors with Posterior Sampling for Reinforcement Learning accepted at ICLR 2024!