We currently can either apply outlier rejection to the reserve stars as we iterate on the solution, or don't do any rejection on the reserve stars. The former is often problematic, because too many reserve stars tend to get removed before the fit stabilizes, so we usually don't enable that. But at the end, it should be more reasonable to apply the outlier rejection to the reserves to get them on the same footing as the main sample in terms of quality.
We currently can either apply outlier rejection to the reserve stars as we iterate on the solution, or don't do any rejection on the reserve stars. The former is often problematic, because too many reserve stars tend to get removed before the fit stabilizes, so we usually don't enable that. But at the end, it should be more reasonable to apply the outlier rejection to the reserves to get them on the same footing as the main sample in terms of quality.