It doesn't even make sense to talk about cases per existing case for imported cases, because by definition they're not caused by existing cases within the count

It doesn't even make sense to talk about cases per existing case for imported cases, because by definition they're not caused by existing cases within the country. Like, maybe that could help tame down the clusters from those imported cases, but then you have to impose lockdown restrictions indefinitely and lose one of the big benefits of elimination over the "try and keep R just below 1.0" strategy - being able to lift those restrictions and avoid the ongoing economic and social damage they cause. Also, I think that in reality this isn't actually as scale-free as the nice simplified models suggest, and having 100x the population would make it harder to stamp out spread within the country. Remember, the number of cases per existing case isn't some fundamental physical constant - it's an average over a bunch of different opportunities for transmission with wildly varying expected numbers of new cases per existing case. 100x the scale means 100x the opportunities for low-probability superspreading events that ruin your day.