pub fn benchmark_cpu_parallelism(
    limit: ExecutionLimit,
    refhw_num_cores: usize
) -> Throughput