"One unfortunate problem with nuke plants is that IIRC you have to have a continuous connection to the grid."
Most modern designs have passive safety mechanisms. Positive action is not needed to lift control rods into place (they drop in) and cooling systems are gravity fed.
One of the critical design features of all recent-build plants is the ability to passively cool the reactor system for the entire meltdown risk period (residual heat dissipation) for the entire danger period _without_ electricity.
That said: Water cooled reactor vessels are intrinsically unsafe (water is a nasty corrosive solvent at high pressure + temp, steam explosions if allowed to vent to atmosphere are nasty and there are always radionuclides dissolved in the mix due to the corrosion mentioned earlier). Whoever thought it was a good idea to use liquid sodium as a coolant needs their head read. Molten lead is marginally safer but a lot harder to remove if it freezes and anything with a graphite core that can get exposed to atmosphere is problematic (sellafield and chernobyl fires)
Molten salt U233/Thorium reactors are the best bet I can think of and would be more so if the graphite matrix can be replaced with something else (this looks to have been cracked). They can't melt down, burn or leak (any leaks will freeze solid before going far) and they are able to burn 99% of what goes in, vs conventional uranium systems 2%. Nor do they need expensive and problematic uranium enrichment plants (operational cost of the USA civil enrichment program is classified information). What they're not good at is producing material for nuclear weapons, which is why the USA gave up on developing them in 1972.
If the transformer explosion had happened at a non-nuclear plant this wouldn't be headline news.