Yes, Fukushima did not kill much. Is it because the nuclear is in itself super safe, or is it because we were lucky? We would need ~100 Fukushima before even having a rough estimation.
I think risk cannot be counted as "number of deaths", but as "capacity of losing control". Cars and cows kill a lot of humans. Yet, their risks are controllable: it is not true that if just few things change or if the circumstances are slightly unlucky, suddenly, they will do 1'000 times more deaths. With this way of thinking, we can understand better why scientists are worried about pandemic rather than common cold, even if before the last pandemic (and maybe during, whatever) common colds was killing more.
And it works with coal too: unpredicted effects of global warming are a way of losing control, and it is why scientists are so adamant about stopping using coal.
I don't think it is true that we have judged the other forms of power generation differently than the way we judge nuclear. We judge the same way: based on our current understanding of how easy it is to keep or lose control.
Sure, coal is dangerous and has unpredicted effects. But the reason it was treated differently is because the capacity of losing control with coal were not at all obvious from the start, while they were written in black and white from the start for nuclear.