There are several ways a spacecraft engine can be efficient. The fundamental thing is the tyranny of the rocket equation: in order to get faster you need fuel, and then you need fuel to push that fuel, into a compounding way. The best way to address this tyranny of the rocket equation is to have an engine that expels something at a very high speed. The best chemical rockets push hot gas at about 4.5 km/s. The engines discussed here are "magnetoplasmadynamic thrusters" [1]. They push plasma with a velocity in excess of 15 km/s. Some studies suggest even 100 km/s. They are basically the most efficient rocket engines our current technology can build.

The next type of efficiency is how much of the source energy is converted into the energy of the plasma jet. Is it 50%, 90%, 99%. That I don't know, and obviously you want as much as possible.

But this project is addressing a different problem. In order to create this plasma jet, you need a very strong magnetic field to contain the plasma while it is being accelerated. The field itself does not provide the propulsion, it is there only for containment. You don't want to expend a lot of energy maintaining this field. Superconducting magnets are much better than regular electromagnets, so it's no surprise that containing this plasma with superconducting magnets requires a fraction of the power of regular magnets. But someone still needed to do the actual research. What works on paper doesn't translate always in things that work in real life. These guys did just that, and now are ready to send a technology demonstrator in space.

[1] https://en.wikipedia.org/wiki/Magnetoplasmadynamic_thruster

You’d get more thrust, but lower delta-v for the same total mass, because you’re replacing propellant mass with fixed mass.

That is an economic question. What does efficiency costs vs what does providing more power cost?