This is all, in the end, due to G.R's inability to allow anyone else to get credit for their work, it seems. I dismiss ENTIRELY any claims made by G.R. regarding the "technical" side of the Enterprise. Nothing he ever came up with on his own made even a lick of sense.The point is that the SFTM was published long after Star Trek was off the air and was not an official part of the production; therefore its contents are not an authoritative source as to the intentions of the actual creators of the show.
And all Roddenberry signing off on it meant is that he saw it as a way to promote his cancelled TV show and maybe make some extra bucks off of it. Later in life, he repudiated a lot of FJ's assertions.
On one hand we should take Gene's word, and on the other we shouldn't? Yes, Gene did a backflip on the Technical Manual, but that was - as you well know - once he had TNG started up and he desired, really, to try to get complete and total control of Trek once again (a situation he, himself, had made impossible). Don't try to conflate and say that it was because Gene got a better understand of Trek's technology... too much in TNG's 1st season says otherwise.
And you know what? There's nothing wrong with that. There's not a damned thing wrong with a storyteller not being technically-proficient, nor is there anything wrong with a technically-proficient person not being much of a storyteller.
The problem arises when you get someone who lacks skill in one area or another, gets other people to do the work, then wants to get 100% credit for what they've done.
The Enterprise was designed by Matt Jefferies, not by Gene Roddenberry... with some obvious influences coming from Datin and crew. Some elements of the design lineage came from various scriptwriters... who put stuff in that, honestly, I doubt Roddenberry had much of a clue about. (Had he really understood all of that, we wouldn't have some of the contradictions we have... after all, Roddenberry was never shy about rewriting anyone's work if the characterizations weren't consistent with his ideas, was he?)
FJ worked from notes which came, in some cases, from Roddenberry, and in some cases from those who actually influenced the design of the ship more directly. And he did a "based upon" fictional work that was NEVER really "a direct representation" of what we saw on-screen. Hell, his little "foreword" messages addressed this directly, by referring to it as being "incomplete" or "edited from the real information to avoid harming the timeline" or whatever... he knew he wasn't consistent, but was trying to make something that would "work" where, as far as he was concerned, certain aspects of the show didn't work.
I love FJ's work, and consider it far better in terms of technical content than Roddenberry's occasional off-the-cuff pronouncements, but NEITHER is what I would consider "source material."
MJ is the principle "source" as far as I'm concerned. And anything seen on-screen has to be used, though very little of what was seen on-screen was REALLY inconsistent (since most of it was fairly vague). Little things which were created, but not readable on-screen (such as the 947' length, or names of ships on a chart) are irrelevant because they were never intended to be seen... and thus fall into the category, as far as I'm concerned, as the "medical insurance" status bar on the TNG sickbay monitors.

MJ intended this thing to be an analog to the radar nosecone on an aircraft. A sensing device, in other words. There was never any clear indication that it was "also" a deflector, so that remains pure speculation... albeit (see above, re: gravity-based deflector-sensor functionality) a reasonable path of speculation.