Why would you do it anyway though? If you can generate the energy to do the reversal... wouldn't it be a lot more efficient just to use that energy directly in propulsion?
As stated, it's an emergency system. The idea is that if a starship runs out of antimatter, it can use fusion energy or perhaps solar power to
very gradually generate enough antimatter to power the warp engines long enough to reach the nearest gas station, err, starbase. It might take days or weeks sitting idle before you have enough for a warp hop of a few hours.
The other thing that always bothered me is that matter-antimatter reactions = high-energy gamma radiation. How do you then render the gamma into usable form?
Electron-positron annihilation produces gamma rays. But proton-antiproton (or neutron-antineutron) annihilation produces both gamma rays and pions, and charged pions quickly decay into muons and neutrinos, with the muons decaying in turn into electrons, neutrinos, and their antiparticles. The gamma rays and neutrinos escape, but those charged pions, muons, electrons, and positrons can be redirected and contained by a magnetic field and used as an exhaust stream or as a means to heat a propellant, according to real-world proposals for antimatter rockets. So these particles probably constitute the "warp plasma" that results from the annihilation reaction and is used to channel energy to the warp coils.
As I understand it, the "magical dilithium crystals" have their own unique magnetic properties that render them "transparent" to the annihilations common to M/AM reactions. When mixing matter and antimatter the antimatter is channeled through the crystals. This is the so called "tuning." Not sure why this is necessary.
The lattice structure of dilithium is such that, when in an EM field, it functions as a sort of "magnetic bottle" on a microscale, holding the antiparticles within the gaps in the atomic lattice so that they don't annihilate the crystal's particles. As for why this is needed, I assume it's to concentrate and confine the reactants. If you just allow particle and antiparticle streams to slam directly into each other, most of the particles will miss (because the streams are mostly empty space), and the eruption of energy from those that do annihilate will blow apart the remaining particles and prevent them from coming into contact, halting the reaction. Presumably, a dilithium crystal channels the particles and antiparticles into colliding more consistently and keeps them confined so they don't get scattered, thereby intensifying the reaction greatly.
However, use just a teeny bit more matter than antimatter in your "blend" and the resulting gamma rays energize that remaining matter which would in fact be the "fancy plasma." Channel that plasma all the way to your warp coils and voila.
That's quite possible -- except it contradicts "Coming of Age," which alleged that the matter/antimatter intermix ratio must always be 1:1. However, your suggestion is closer to real proposals for antimatter power -- using more matter than antimatter and using the heated excess matter as a working fluid or reaction mass -- so I'm inclined to disregard what "Coming of Age" claimed. In fact, that would probably be entirely necessary to deal with the problem of gamma-ray escape mentioned above. Those gamma rays would represent a lot of energy, and it would be kind of pointless to let them just radiate away uselessly. If they instead irradiate the extra particles in the matter stream, those particles will absorb much of their energy and be able to transmit it to the engines.
So I'd say you're right. The pions, muons, electrons, and positrons resulting from the deuteron-antideuteron annihilations must only be part of the warp plasma, with the rest being extra deuterium nuclei energized by an absorption of gamma rays (which might split them into protons and neutrons).