Agreed, and ditto if I forgot to welcome you, Egger.
A few assumptions/decisions on my part with regards to the evolutionary path here. ...
Here's my theory of warp drive and the reactor configuration in TOS:
1. The Constitution class has 3 reactors, one in each nacelle and one of the same type in the engineering hull.
2. Dilithium (used in the "Dilithium Crystal Converter Assembly) is a highly efficient way of converting massive amounts of energy in the form of radiation into electrical energy.
3. The "Dilithium Crystal Converter Assembly" is the so-called "main energizer", with auxillary (non-dilithium) energizers for example being used for the impulse reactors.
As we have seen in Star Trek Enterprise the matter/antimatter reactor can be placed in the hull, so danger from the reactor is no reason to place them outside the hull (in the nacelles).
I think, as the faster and faster warp drives after Enterprise needed bigger and bigger reactors, there was the choice of one huge reactor in the hull taking away part of the volume of that hull, or multiple reactors. The reason for their placement in the nacelles is for convenience. As I see it, the warp coils need radiation energy, and therefore there is no need for a Dilithium Crystal Converter Assembly in the nacelles. (see NOTES below)
For the other systems besides the warp coils, there is a small non-dilithium converter assemby fitted to the nacelle that is similar (or even identical) to the ones connected to the Impulse reactors. These can convert only a fraction of the energy coming from the matter/antimatter reaction to electricity, but it is enough to power the systems in the nacelle.
With this configuration, a nacelle is a complete energy production and warp drive system that only has to be fitted to a starship hull. This hull contains a Dilithium Crystal Converter Assembly (main energizer) which is connected to the reactor in the nacelle and converts as much energy into electricity as is needed to power the other systems of the ship. Also, this energizer diverts some of the warp plasma to the EPS system as a transport medium for the electricity.
Now you could ask: If the warp engines don't need dilithium, why is the warp drive not working when the crystals aren't working?
The answer is that a ship needs more than just the engines to use warp drive. Without the deflector(s), the tiniest bit of space dust would blow a hole into the hull.
So even if the warp engines themselves work, you can't risk going to warp.
Now on to the reason for a third reactor:
A ship like the Saladin class for example has one nacelle and only one reactor within it. That means that:
1. The ship is rather slow because it has only one set of warp coils.
2. The energy that the one reactor produces must be split between the engine and the rest of the vital systems (especially the deflector(s))
3. When it uses its warp drive, shields and weapons cannot be operated with full power (if at all).
So that ship (or, more precisely, a ship that has only nacelle reactors) has no energy left for other systems when it uses its warp drive (at least with high warp factors). That is why the Constitution class (as the best Starfleet has to offer) has an additional reactor. With it, the ship can give every system full power at the same time.
Now on to TMP:
As I see it, the Intermix shaft IS the reactor (so from there on: no nacelle reactors anymore). This is a new design which is easy to adjust to the different energy needs for ships of different sizes, since you only need to make the shaft longer for more energy or shorter for less energy. For example, a refit Saladin class would only have a short vertical shaft from the deflection crystal down to the nacelle, and that is enough for that ship class. Bigger ships have longer shafts with the Excelsior having the longest. There still is a Dilithium Crystal Converter Assembly needed for this configuration, as seen in TWOK. By the way, when the main energizer in that movie failed, Scotty then rigged all auxillary energizers to convert enough energy from the intermix shaft to power the ship. This bypass was so unstable (like a chrismas tree) because the auxillary energizers were not disigned to handle that much energy.
And on to TNG:
By the time the TNG-style core came into use (whenever that is) the Dilithium Crystal Converter Assembly had been incorporated into the matter/antimatter reactor itself. There, it converts part of the energy into electricity for use in the ships systems. That energy is tapped off from the warp plasma conduits by the main EPS taps and, together with some plasma, is then fed into the EPS system.
This stems from the observation that, as seen in TNG, plasma is pumped up to the nacelles and "injected" into the warp coils. If the coils use electricity, I think there would be no reason to "inject" it there beacause the plasma would only be the medium for the moving electrical energy (for comparison: you don't "pump" power cables into your computer, you use them to "pump" electrical energy into it). Also it wouldn't be "highly energetic" since this energy would be already converted to electricity.
Additionaly, if the warp coils would use electrical energy, the placement of the reactors in the nacelles
and the energizer system in the secondary hull
would make no sense. Plasma would be pumped down the pylons to the energizer and then up again to the warp coils.
And there cannot be additional "Dilithium Crystal Converter Assemblies" in the nacelles because in the series, the failure of the crystals in main engineering always resulted in the ship losing main power.
Phew! I hope I didn't forget anything.