The dectector and the variable field (that induces the localized measurable changes) stop between scans, but the static magnetic field is kept up.

As long as you keep up the superconductitvity there is basically no electrical loss in the coils. Dialing the magnetic field down would require pulling out the energy, and reinjecting new energy to get the field back up. That’s the slow part, because injecting current quickly would heat the coil above superconductivity, leading to a quench.

I’m not sure how energy is withdrawn in the ordinary shutdown procedure, but I expect it is exchanged into heat and vented to the outside air in some way, rather than reinjected into the grid in a usable form. (The latter would require an inverter to turn the DC back into AC synchronized to the grid, probably would increase complexity by too much). So I suspect it would be wasteful too.