Optimal Long-term Contracting with Learning