While it's always good to keep one foot in the real world in discussions like this, and keep in mind what the writers may have been thinking (or not as the case may be) at any given time, I think such speculation should take a back seat to actual clear references in the show itself.
We have several specific references to "anti-matter nacelles" in TOS and TAS, and if we remember that the term "anti-matter pods" was often used interchangeably with "anti-matter nacelle" (sometimes in the same sentence/discussion) then the number of references goes up considerably. And it does no good to point out that the term "pods" in later "TNG era" trek referred to something else, because it had nothing to do with TOS/TAS.
Now, where does this leave us? Well, we need not assume that anti-matter is stored in the nacelles, however likely that may seem (to some), but clearly, matter/anti-matter is in the nacelles doing something for some reason; presumeably, powering the warp drive.
Thinking outside the box for a moment, has it ever been considered that perhaps the main dilithium/reactor "core" actually creates the anti-matter that is fed to the nacelles, where it is then utilized or possibly stored? Is there anything in TOS/TAS to contradict this possibility? In other words maybe we've had it backwards, maybe dilithium doesn't so much moderate the reaction of M/A-M reactions as it does create anti-matter by converting matter fuel in some way? Then the resulting A-M products could be sent to the nacelles to mix with matter, and the resulting energy used to power the warp drive (coils)?
Such a design would not necessarily be inconsistant with later trek tech either? We know in the real world the products of M/A-M reaction is roughly 50/50, half being gamma rays and neutrinos and half being electrons and positrons. Perhaps then in (TNG) trek, an initial M/A-M reaction, "moderated" by dilithium occurs, in which the 50% consisting of gamma rays/neutrinos are used to power the ships systems, and the 50% consisting of electrons/positrons are seperated and fed to the nacelles as the "warp plasma" we've heard about?
So perhaps the "plasma injectors" in the nacelles then, are two of a kind, electron injectors and positron injectors (hence M/A-M nacelles), and they introduce this M/A-M in the open cavity we've seen, where the resulting anihilations are mostly contained by the warp field, but allowing energy absorption to the coils? Perhaps the coils are initially charged to a low level operating capacity by fusion reactors, also presumably in the nacelles, and then "ratcheted up" with M/A-M energy for successive warp factors?
Sorry for the long ramble, I just put this all together, so haven't thought all this through completly, so I may be overlooking something, but I thought I'd throw it out there for everyone to ruminate over.