Fix the GUARD controller event
As discussed in OnionPerf's #33421, the
GUARD controller event is mostly broken.
mikeperry thinks it might have happened as part of proposal 271.
A quick grep for
control_event_guard in 0.2.9 and current master shows that this function is only called in a single place today when it was called in several places back in 0.2.9.
asn kindly offered to bring this event back with an estimate of 4 hours. Thanks!
This would help us a lot in OnionPerf, because we could then better track which guard is used for measurements.
If it helps for testing, I can easily spin up an OnionPerf instance that runs a feature branch, let it run for a day, and post the controller event log here.