GenerateAndCheckPerfResults currently allows only a hardcoded subset of users to override a failing instruction-count check on benchmarks_sep. When a legitimate PR introduces a new call along a measured path (for example findOneById) the instruction count regression blocks merge, and only those listed users can push it through. The list should be expanded to cover at least all of Product Performance, and optionally a small set of server engineers who regularly land perf-sensitive changes.

Preferred implementation: move the list out of hardcoded Python into a data-driven source (config file or Jira group lookup) so adding someone does not require a code change and review.

Additional Context

• DEVPROD-30962 Improve SEP benchmark reliability: Container for SEP benchmark reliability work.

Ticket originally authored by alice.doherty@mongodb.com, edited by james.littlejohn@mongodb.com during triage.