If you step outside during the day you can have first-hand experience of a contained fusion reactor that manages to heat things up at a distance.

That's how we'll extract energy from fusion reactors, but with approximately 1 AU less distance.

( unless things go comically wrong )

> and the shell is super-cooled to 4K

I think only the shells of the superconducting coils are cooled to those temperatures. The rest of the system, including the containment walls, are water-cooled.

In a real system you would also have neutron radiation carrying away energy that could be captured with additional layers of water.

The plasma is confined but its glow is not.

i have absolutely no authority to speculate, but i would guess you'd vent plasma to a heat exchanger by diverting only part of the stream with a controlled EM field, either periodically or continuously. You'd have to do it anyway to remove the helium and inject some fresh tritium and deuterium, right?