Ghost offline runners accumulate when spot instances are terminated

GitHub runner registrations remain in "offline" state after EC2 instances are terminated (e.g., spot interruption). The scale-down Lambda only iterates over existing EC2 instances to find orphaned runners, but doesn't check GitHub runners for missing EC2 instances. This causes offline runners to accumulate over time. any ideas about it?