No-Regret Exploration in Goal-Oriented Reinforcement Learning