That more than the declared 456 are infected is not in doubt. But beyond that, it's very difficult to predict and would need data and statistical modelling tools beyond my capabilities. You can have a stab at it by looking at how "random" outbreaks are. If they are very evenly spread throughout the UK, then that implies widespread infection in large numbers. If it's more "patchy" then very likely the numbers are fewer. A bit like tossing a coin 10 times = you might get 8 heads and 2 tails. But toss it 1,000 times and there is no way you're getting 800 and 200. It will be more like 532 - 468. 100,000 times and maybe 50,320 vs 49,680. The more cases, the more predictable the outcome.
So my gut reaction is that we may have 2x, 3x, even 5x the number of cases, as yet untested. I hope so. But I doubt "tens of thousands".