Probably for the same reason that I've seen processes put in place for performance testing a service (internally or on production) - because it puts a significant load on the service and automatic rate limiting, flagging, etc can be triggered.