Variance in benchmark results: review hooks.py once we have more data

[slobentanzer]

For many of the initial benchmarks, we only have mean values and iterations. Once we have enough additional material, we can remove the code that deals with this distinction and go fully to the individual recorded results and summary statistics computation.