Why not both? Complementing explanations with uncertainty, and the role of self-confidence in Human-AI collaboration