Occam's Razor my fiend. Ask yourself what's the higher statistical probability in Trump's America:
A minimum wage teenager, in a city of less than 50,000, on the tills during the breakfast rush, managed to positively ID the suspect, despite him wearing a mask and beanie, based on ambiguous CCTV photos, and decided to snitch. He was then detained within 11 minutes.
Versus
A federal agency tapped into CCTV or kiosk-cameras automatically ID'd him via facial recognition, and then established clean, admissible evidentiary trail to mask an unconstitutional investigative source.