Predicting Software Defects in Hybrid MPI and OpenMP Parallel Programs Using Machine Learning