Issue with reproducibility using recompute() after gs_pop_add()

I am getting different cell counts every time I run the same script where I use `recompute()` after adding new gates using `gs_pop_add()` that are boolean combinations of preexisting gates. This has affected downstream analyses where I look for significant differences between subpopulation frequencies. How do I ensure that the gating is reproducible and consistent?  