Although I can understand why they would want to sell it as a battery, that is not a real accurrate description of the device. A nuclear powered fuel cell would be a more accurate description.
"Battery" is any multiple-cell connection. It is most commonly used to refer to batteries of electrochemical cells (and has since come to be commonly used even for a single electrochemical cell, which is NOT a battery.) If they hook more than one of the silicon cells together "battery" would be a proper term for it. (Of course it would equally be a proper term for a solar panel. B-) )
I think they're using the more common meaning of "battery" as a self-contained power source.
As to being more closely analogous to a fuel cell I can't agree. Fuel cells have a fuel input and an exhaust output, and otherwise are much like electrochemical cells. This thing is nothing like that. (And you REALLY don't want to be refuling it with radioactive gas at home.)
It's actually more closely analogous to a solar cell - in a box with "a little bit of sun". B-) It's a diode, collecting electron-hole pairs generated by externally applied energy bombardment - though in this case the bombardment is fast electons, each creating an burst of electron-hole pairs, rather than photons creating a single pair each.
= = = =
I've wondered for decades - since TI came out with a tritium dial watch in the '70s or so (LCD quartz watch illuminated by a tritium-filled phosphor-coated can for a backlight) - why they didn't just stuff some chunks of solar cell material in the can with the tritium to power the watch. A quartz watch and its tiny LCD panel requre vanishingly small amounts of power.
I guess now we know: The volume of p/n junction region would have been too small to pick up enough power to do the job - and they finally worked around that by texturing the surface to increase the junction volume and its exposure to the fast electrons from the radioactive gas.
By the way: This should work with alpha particle emitters, too.