EARS: An Anechoic Fullband Speech Dataset Benchmarked for Speech Enhancement and Dereverberation