Add 28 benchmark datasets with rich visualization views

Replace local DREval with HuggingFace REval adapter

Rebrand to ML4SE Benchmark Viewer, make datasets a core dependency