Analyzing Resource Utilization in an HPC System: A Case Study of NERSC’s Perlmutter