The First Comprehensive Dataset with Multiple Distortion Types for Visual Just-Noticeable Differences