On the Design and Evaluation of Human-centered Explainable AI Systems: A Systematic Review and Taxonomy