Add self-comparison stability testing

[fbusato]

Evaluate a give benchmark multiple times in a stable environment/system to track SAME, FAST, SLOW, and ambiguous outcomes. The evaluation focuses on maximizing the number of SAME results.