The power failure is not "low probability", it is the dominant failure mode that happens somewhere around 1 in 1000 to 1 in 10,000 reactor years.

Reactors were licensed in the 1970s based on an entirely wrong model which saw the dominant failure mode being the pressure vessel bursting. Laymen have a totally wrong point of view about that, they think a pressure cooker really has the metal burst and go off like a bomb, really the seal breaks and you get sprayed with superheated steam which is dangerous enough. Pressure vessels burst because the chemicals eat them from the inside out but for every pressure vessel that bursts thousands of storage tanks get sucked in.

After TMI the model was updated to recognize "station blackout" as the #1 risk.

Fukushima currently has a body count of 1 and the city is perfectly habitable.

It was also basically the worst case scenario that could happen to that reactor design.

The tsunami and the earthquake killed 20000 for a scale.

Sure, I think that's true, but it was an event of such magnitude that there's a clear state interest in regulating against it! They had to evacuate a huge area around the disaster, for some time. Sure, it was a best-case scenario, but it was the best case of a worst case.

(I support nuclear power, for whatever that's worth. I think it's a good idea and we should do a lot more of it.)