Mind the Uncertainty: Risk-Aware and Actively Exploring Model-Based Reinforcement Learning