Coz the paper gives a function for extrapolating from these tests. This is purely testing thermal decay.

10,000 years sounds like a good benchmark and isn't as obviously ridiculous as saying a million years at 260°C