Calmness and Exact Penalization in Constrained Scalar Set-Valued Optimization