Welcome to the community Mark.
It’s very subjective, and with all things actual pain levels from users will ultimately tell you what you need to aim for. I assume you are talking about backgrounded, not foreground?
On IOS, backgrounded apps don’t use power, but suffer from loosing state/stale and drop connections and have to re-initialize. android apps are easier on backgrounding, but do use power. Foregrounding gets more complicated as in the thread, because screen brightness and size completely rule your power consumption on any platform. You can get companies that will retrofit an android with a special accurate power monitor, the inbuilt monitor is not very accurate anyway.
I would grab a typical device and actually use it as a user for a day and take readings and not rely on the device stats as much as on user experience. It’s really dependant on what the app does, and on who your user is, and what devices they typically use. Even so, it’s pain managed in the end. That’s how we do it where I work. Google and Apple have verification “programs” that may also trip you up when they test your app and it turns out to be a beast.