Should Most Overpowered Supporting Cohort - Crafter or Buff-Station? remain open, or be closed?
It should be closed.
I initially voted to close this question, but then I retracted that vote and answered it based on the precise wording of the question at the time, with a fairly tongue-in-cheek answer that was supposed to illustrate the idea that, that tongue-in-cheek answer aside, there is no answer to the question.
That was clearly a mistake. I should have just left my close vote.
There are many, many reasons why this question is impossible to fit into the Stack Exchange format:
Constraints are a “no true scotsman” situation
The question has been edited multiple times in response to answers suggesting solutions, to in effect say “well no, not that,” and we still have no idea what is or isn’t actually acceptable in an answer. The question alludes to a pair of apparently-highly-specific builds that must be considered precisely in any acceptable answers, but doesn’t indicate the details of either build. (That pair of builds also constitute a clear false dichotomy with respect to the title question, but we’ll ignore that I suppose.)
The OP’s edits and comments have referred multiple times to “natural” parts of the game, “standard optimization techniques” versus that which “defies the purpose of playing the game in the first place.” There are no “standard optimization techniques,” every table is different in what is considered acceptable and what is not. The things the OP considers “natural,” would not be allowed at any table I have ever so much as heard of—which is fine, whatever floats OP’s boat, but it is necessary for them to indicate what it is that floats their particular boat. OP seems to be under the misconception that everyone ever is on the same page with respect to things that vary wildly from campaign to campaign.
Goals are completely unspecified
Optimization questions on this site are required to be very general—looking for best practices, things to be aware of, etc., for general character concepts or archetypes—or else highly specific, indicating exactly what metric or quality is being optimized.
This question just wants to know the “best,” with zero definition of what that would entail. This is, quite simply, precisely the kind of thing we have decided we quite definitively cannot handle:
The common thread here is that the asker provides answerable requirements. As Brian says below if a thread does not have answerable reqs then we should be voting to close until they provide them.
Even if all of that was specified perfectly, the question is still vastly too broad
Per comments, the request is for an answer to
Select spells for buffs, select items for craft, prove what results in better numbers and why.
Quite simply, you would never be able to do this comprehensively or definitively within the character limits of a Stack Exchange answer.
There are thousands of items, and thousands of spells, in the system. Any given character can trivially have access to hundreds of each—particularly with the kinds of high-power builds that the OP alludes to. Even with a clear goal and clear constraints, that means a prepostrous number of combinations to consider. This is the knapsack problem, which “has been studied for more than a century, with early works dating as far back as 1897,” due to its difficulty.
And beyond that, the answer is going to be highly-specific to one individual in one campaign
If we nail down all of the specifics of all of the builds and all of the context in which decisions are made, we have narrowed things down so far that the results are going to be meaningless to anyone aside from that one PC seeking a cohort. Even small changes in any of the parameters could result in wildly differing results, because what is “best” necessarily relies upon what you want and/or need.
It’s been said before, but worth repeating: SE is not a good place for novel optimization
Stack Exchange is an excellent place to record and explain optimization that has been achieved elsewhere, but it’s a terrible format for actually developing optimization. The aforementioned knapsack problem applies, really, to most or perhaps even all RPG optimization problems, and the only way to achieve even a reasonable approximation of a solution (given that the knapsack problem is, for practical purposes, unsolvable with the number of potential inputs available in an RPG) is to have a lot of people build off of each other’s work and/or head out into the wilderness to consider entirely new approaches, and iteratively build up a best approach. We can’t do that here; answers are supposed to be comprehensive, definitive, and largely fixed, not constantly updated with new contributions and ideas from myriad users and comments.
In general, this isn’t necessarily a reason to close a question—for an optimization problem that just hasn’t seen a lot of attention, it is usually better to just leave it unanswered, or (if there’s been a little bit of work done on it) to answer with the current state-of-the-art and note the limitations of such an answer. But considering all of the issues above, I think there is no merit to keeping a question we will never be able to answer open.
WHY THE QUESTION SHOULD BE OPENED AND ANSWERED "AS IS":
"This GM would not allow this at his table." Game and rules should be discussed, commented, analyzed in complete separation from "this GM" or "that table". If GM has a say in what is or isn't acceptable, this is no longer a RAW discussion but a houserule discussion. Whether that's a realistic assumption when it comes to real game table is completely irrelevant.
On a personal note though, my own experience (both as player and GM) says that Players and GM don't step on each other's turf. GM might throw a Template of Perfection on everything to be able to challenge players but he/she has no right to say a word about how far PC optimization goes. That's players department, not GM's, GM's right is to adjust the content to the new "actual power level" of his players.
The only thing that I'm used to commonly being "restricted" are things that break the game (in a programming definition, as in "crash", not as in "too strong"). Those would be infinite loops, infinite wishes, pun-pun or things that completely remove entire feature of the RPG system from the game (for example using Thought Bottle to double the amount of spells per day for basically free, at least compared to Pearls of Wisdom or Rings of Wizardry is completely fine, because it's still limited, while using the same Thought Bottle to generate infinite amount of crafting/casting XP is a no-go). In a similar matter using free Wishes (bound Efreeti) is fine as long as used to replicate any spell (effectively having access to any spell in the game but still limited to 3 uses per day) but using the same free Wishes to create infinite stream of GP is a no-go. All common sense, no need in having to explain this in every question.
"What campaign are You playing?" Another thing that is completely irrelevant to RAW solutions. If the goal is to optimize some aspect of the character, the answer should be a ready-to-use optimized character. Whether it's actually a playable character in real game table is completely irrelevant to the question, unless specified otherwise.
Ultimately RPG is just a numbers game, statistics to be specific. Optimization can be done without ever needing to actually play the game, thus specifics related to the actual game doesn't matter. System itself provides tables for difficulty of pretty much everything, from combat encounters to skill checks. Optimizing means gaming those tables, going above and beyond what system designers deemed "challenging" for a specific PC level.
"Too much to analyze within SE character limit..." Not exactly. Majority (99%?) of items and spells are completely and utterly useless. They are at best fluff-fillers or situational toys. As far as powergaming goes, there's only a small pool of usable things that just work regardless of campaign or character details.
"Unspecified definition of "the best" in the question. Maybe this is my video games background but "the best" is very specific for me. Something that can break through every content designed for it's power level or higher (in case of D&D that would be a Character Level vs Challenge Rating). So for example it's obvious that character who walks through a CR balanced for him +3 is stronger than a character who walks through a CR balanced for him +1. Pure numbers, simply treat RPG like a video game, and everything becomes clear.
Ultimately "the best" RAW character, would be one that has 100% success chance on every roll possible, against all the official content ever created (yes, including epic, as most people tend to forget existence of this). That is of course impossible to achieve without exploits but it is a good measure for what the goal is. I actually simplified this a lot by restricting answer to Fighter character, which means only combat is relevant.
"Novel optimization thing." While I agree that novel optimization is "on the edge" (doesn't cross this edge though) when it comes to being suited for SE, it's also probably the only place with enough community where it can be attempted, which justifies going a step further. The only real difference is that it requires a lot more from the person answering the question. It's not longer a "one paragraph for free SE points" or "answering someone who is too lazy to check rulebook" but becomes a "use Your character limit well and do quite a bit of research before answering".