Keep in mind that most of that comes from MHD simulation and MHD simulation are well known in the community to get details of reconnection quite wrong [1]. So until somebody has confirmed that with particle-in-cell simulations, hybrid simulations with a good Ohms law or an experimental device, take all performance numbers with a big grain of salt.

Oh an another thing: Very high velocity exhaust gives you large I_sp, so good efficiency in the use of the working gas that you are exhausting to get delta-v. But high exhaust velocity also implies that you need A LOT of (electrical) energy to do that. (Since power requirements scale as P = 0.5 * m/dt * v_ex^2, but thrust only goes up linearly, i.e. F = m/dt * v_ex.)

[1] See figure 1 in https://agupubs.onlinelibrary.wiley.com/doi/10.1029/1999JA90...