This article was a lucky find. I've already forgotten where/how. Glad you liked it :). And plasma is very fluid-like, usually (see: magnetohydrodynamics/MHD). But also, here, relativistic. And collisional. And radioactive (OK, that one's kind of a stretch).
Btw, the simple-ish case of the Kelvin-Helmholtz single-fluid flow shear instability is popping off almost all of the time, like 15 Earth radii away, along the duskwards magnetopause flank. 100% MHD plasma. And then rx'n (ignore this for now).
Sorry, I should've gone into more detail, but now you've got it. One main function served by the neutral beams is to heat the plasma via collisions (in tandem with the B-field), and the plasma then hopefully makes it past some critical pressure and collision frequency threshold to self-sustain and at least temporarily generate net positive energy. I would say heating via neutral beam is "the" main function, but again, H-mode.
In that first configuration of the "four ways to start an FRC" cartoon in the article, the neutral beam is injected along the field. So that leads to disproportionately parallel heating of the plasma, which makes a localized (in space) anisotropy in ion (and electrons, to another degree) velocity space, which always wants to find a way to be thermalized, i.e. made fairly Maxwellian again, instead of having two (or more), or even nearly two distinct populations. In the collisionless regime, this happens via wave-particle interactions, but for fusion plasmas, it is almost always the collisions that serve as the interactions thermalizing the plasma.
So uhhh anyway. The article talks about a totally different type of instability. There are often many competing instabilities, it's kind of like a race to see if good conditions are sustained for one instability's progression into the runaway, non-linear growth phase, and then that instability can go on to disrupt and/or initiate other processes. In the article's case, it's a disruption like a vortical, fluid-like eddy current, a not-quite-chamber-size perturbation in the magnetic field geometry. In most tokamak losses of confinement, Princeton, France, et al. seem to think a sustained topological resonance can enable another instability, the tearing (or plasmoid) mode, to progress into the non-linear phase and then trigger magnetic reconnection (rx'n), which rapidly breaks confinement stability. Rx'n is by far and away the most important non-linear plasma mixing process, and I suspect that rx'n is to blame for almost every significant loss of confinement.
Rx'n seems typically required for rapid system-scale topological reconfigurations. We have found rx'n to be a crucial mechanism coupling the turbulent cascade of energy from large to small scales over time, in the form of electron-only rx'n resolving ion-scale turbulence; pockets of magnetic domains in a plasma, separated by ~10 ion gyroradii or ~10 ion inertial skin depths (somethin' like that, depends) with some significant component of each's magnetic field opposing the other can annihilate some magnetic flux along the axis of component reversal, converting the flux into electron energy (and not ion energy, since it's not ion-coupled).
Electron-scale rx'n recently solved the coronal heating problem, yo! Big coronal mass ejections and strong flares are much more ion-coupled, though. Basically all of the mass of classical plasmas is ions, so yeah.
I shall cut me off here
last edit, swear on my eldest cat's life
eh, oh well, she was sacrificed b/c I found the .pdf