Why is a long-running scheduled search running multiple times over a short time period?

[esalesapns2]

Splunk Enterprise, v7.0.3

I ran the search in https://answers.splunk.com/answers/750097/search-performance-impact-how-to-find-user-deployi.html

I see the exact same alert running 9 times in a 5-minute period that eats 130 to 165 data.pct_cpu each time.

It's an Alert private to a user scheduled to run at 8:00. It ran 9 different times from 9:23 to 9:28 am, according to the search.

Is it a bug in the search due to the way "stats latest(data.pct_cpu) is used that makes the "_time" off?