Minimax Optimal Variance-Aware Regret Bounds for Multinomial Logistic MDPs | Steady Practice | SteadyPractice