As a common thing one-shots - for or against the PCs - are a bad idea once beyond the very low levels. As an occasional spectacular exception that the players will talk about for years, though? Sure! Bring it on!
Lan-"and this right here is the base rationale behind every critical hit system ever made"-efan
Again, it depends on the system. Savage Worlds makes virtually everyone except for Wild Cards a one shot kill. And there are several systems that don't use HP. And, even as a Wild Card, you get stunned pretty often by a hit. In other words, combat is far more dangerous than it is in D&D. But, since D&D uses HP, combat generally isn't as dangerous as it is in other systems, so, we tend to play a LOT more combat in D&D than in other systems. Which, honestly is pretty darn fun too.
Just so I understand correctly, the power that you want a high level Fighter to have, the thing that will give them some kind of parity with the Magic-User, is the ability of holding the Magic-Users stuff in the forge while the Magic-User creates a Magical Item.
And this is the huge proud nail that you want hammered?
Obviously I can not speak for everyone but yeah as far as I am concerned the Fighter can hold as much of the Magic-Users stuff as they want.
Why is it "magic user stuff" to craft a magic hammer? Other than tradition, there's no real compelling reason to say that a fighter can't make his own magic weapon. The last Avengers movie showed exactly what it could look like with Thor essentially holding his hand (well his whole body) in the forge so they can make the new weapon. Crafting a magic weapon is only "magic user stuff" because that's the way it's always been.
And, well, I'd say you missed the larger point. We're arguing about whether or not a fighter should be able to do stuff at very high levels that a caster can do at much lower levels. In 5e, can your 15 strength character jump 20 feet using skills? Maybe. Depends on the DM and the dice. Oh, you're a caster? Go right ahead, no problems, no DM adjudication and no chance of failure (Jump spell, fly, and I'm sure there are other options). For some, this is a pretty large issue because it strongly narrows the narrative space that you can use unless you play a caster.
Of course, 5e solves this problem by making pretty much everyone a caster, so, that's one solution.
--------
And, just a point about tiers of play. It's not like 4e invented this out of nothing. The tiers of play are simply a recognition of how the game is presented in every previous edition. It's not like you take your 1st level characters into the Demonweb nor do you take your 15th level characters into the Caves of Chaos. Can you? Sure, but, 4e was pretty much on target here making explicit what was implicit in every other edition.
So, if the game virtually always follows that same path - zero to superhero - shouldn't the mechanics reflect that? A 1st level AD&D fighter is just as capable out of combat as a 15th level 1e fighter because 1e didn't really have any skill system. 2e added a skill system, but, since it was stat based and not level based, the effect was, by and large, the same. Your 1st level fighter and your 15th level fighter didn't really have much of a difference out of combat - they could talk just as well, hide just as well, jump just as far, ride a horse just as well. Virtually no difference.
3e started the path by making skills level based, but, it also wound up with the in-game problem that skilled characters would auto-succeed and non-skilled characters would auto-fail because the DC's scaled by level. Sometimes. But, 3e had the issue that some skills didn't scale at all, so, it became trivial to auto-succeed some skills. You have a +14 acrobatics means you can automatically move through enemy spaces, for example. Funnily enough, 3e fighters because very effective out of combat because of this because they could leverage their feats into skills and beat the scaling DC's routinely. 4e tried to mitigate the pass/fail problems by adding in leveling to all DC's but, again, you wind up with the treadmill and unfortunately, it was still quite possible to game the system and auto-succeed some skills.
5e uses static DC's that aren't supposed to scale. But, the scale is entirely dependent on the DM and most DM's are notoriously bad at calculating odds. A DC 20 is a HARD DC. This is reserved for things that most people will fail at most of the time. Which means you shouldn't see it all that often. DC 25 and higher should be seen once or twice in a campaign - these are legendary checks like scaling a waterfall or knocking down a castle door.
The problem is, many DM's don't see it that way. OH, you have a 60% chance of success? That's too easy. We'll fix that by jacking up the DC so now it's a "challenge". And, the non-caster classes get screwed over because it always comes down to the DM. If I cast a spell, I succeed. All the time. No chance of failure, no DM adjudication. But, if I don't play a caster and I want to do the exact same thing? Oops, sorry, nope, we're playing a "realistic" campaign.