in AF-19 we saw a system under load with update queries spending time in the index bound building code for $in without checking for interrupt. These updates were holding the RSTL lock during a stepdown and took too long to reach interrupt checks in the execution layer. SERVER-89583 added some interrupt checks during plan enumeration to create earlier opportunities to check for interrupt so that an overloaded node is more likely to be able to kill updates and step down gracefully.

This ticket should consider other places in the query planing code where interrupt checks would make sense and potentially rework the existing interrupt check to be directly in the $in index bound building code