How do you test a bot against one opponent type?

GTO Wizard just showed how to measure one weakness at a time: hold the opponent's range fixed and compare results across all 1,755 strategically distinct flops. https://blog.gtowizard.com/how-to-exploit-capped-ranges-with-custom-aggregated-reports/

How do you check that your bot keeps exploiting, say, a tight opponent across many boards, and not just on a few lucky hands? Fixed test opponents, live matches, something else?