Review single cpu measurement accross multiple rounds/iterations during benchmark
We are measuring cpu time only once for a whole execution of the benchmarking function, what includes multiple rounds and iterations of the benchmarked function. We should review whether this is a good approach and what are the alternatives.