Lifetime of the product. Given such constraints I would seriously consider just starting different projects and let that app die a slow, painful death by having the ecosystem around it move forward.
Yes, the app will be functional on Android 8 in 2039 - if you can still find a functional Android 8 device then (after the epoch roll-over).
Basic maintenance helps the app move forward to Android 30 (or whatever is current then), letting you enjoy precisely the features you've bought back in 2017, so to me that seems to be strictly better.
(I exaggerated version numbers and dates to take in the epoch roll-over issue as an outside-anybody's-responsibility issue with an ancient base OS: https://www.quora.com/What-will-happen-to-a-64-bit-Android-d..., but the same scenario also applies when talking about supporting whatever base OS of 2022 and the ways it will or will not be compatible to today's apps)
What other app developers did in similar situations (seen a couple of time with transitions from free to paid): create a new app in the same space but with the new (hopefully more sustainable) terms, stop providing new licenses to the old app, put the old app the back burner (basic maintenance only, no feature development), hope that the business transitions over.
That upholds the promise (you still get whatever you paid for at some point) better than the other legal alternative (shutting down the app entirely, which ends the "lifetime"). If you want to get rid of the dead weight, and it doesn't happen organically, periodically offer specials for customers of the old app to transition to the new one.
What you should never do: take something that is "lifetime" and make it expire. That only demonstrates that you're not trustworthy.
The idea isn't to capture the entire billion dollar market, but to navigate a market that is large enough to safely feed you and many more for the next decade+, precisely because it isn't scalable (so there won't be any FAANG-or-whatever coming along, eating your lunch)
The 10.1" seem carefully chosen so that on Android, most tablets and phones are covered by the "free" edition while Chromebooks (that can run Android apps) are not.
The pglaf document isn't the most accurate in general (eg. claiming that Germany has "life + 75" terms, while correctly calculating life + 70 right in the next sentence; or discussing that the 56 years rule will end in the US in 2034, since books published after 1978 are subject to life + 70 as well - I guess Gutenberg won't be able to publish a single new book between 2034 and 2048), so I guess the claims made were only about damages (even if pglaf might consider them "punitive") and fines.