i also was just able to split out the criterion benchmarks into their own commit before the other changes so the performance can be compared directly