FIFO: finish={'A': 2, 'B': 3, 'C': 4} mean_slowdown=2.000 reward=-6.000
scan: finish={'A': 2, 'B': 3, 'C': 2} mean_slowdown=1.667 reward=-5.000
shortest: finish={'A': 3, 'B': 1, 'C': 2} mean_slowdown=1.167 reward=-3.500
Capacity sweep checked all six scan orders at each of 16 CPU/memory combinations.
Every schedule obeys both capacity constraints and exactly reconciles cumulative reward with total slowdown.
