The problem of incoherency is manifest in the way magic works (or doesn't). It isn't about outcomes, it's about having two distinct processes (d20 and the whole spell slot mess), and trying to merge them together. Skill-based magic completely fixes the issue, but we're clearly not there yet, with D&D.
Or, to go the other route, you're essentially claiming that merging spells with a new category of nonmagical spells fixes it.
That is part of it, yes. But that isn't the whole score. Depending on the edition, another major problem might be the prohibitive nature of the action economy disallowing anything resembling a balanced, and functional, simultaneous usage of sword and spell. Another problem might be, due to the multiclassing system, the default stats of the character end up not being equal to the task of a front line melee combatant. Another issue (and not a small one) is the crippling effect on overall effectiveness of MAD. Yet another problem, and not a small one, is (depending on edition) there may not be any thematic functionality that actually supports the archetype to make it distinctive; its just a pile of stuff. Here is a good example of these things in action:
In 1992, the infamous, and much decried, Complete Book of Elves was released for AD&D 2e. Within it was the Bladesinger kit. Much like when the Monk was initially released in 3e, people went OMG OP (!) when they saw the kit and considered those bonuses in a vacuum. Those bonuses were (a) + 1 to hit and damage with one weapon (while losing the bonus with the bow), (b) a bonus to defense when spellcasting but an increase in cast time (more probable attacks against you during spellcasting but less likely to get through total), and (c) a + 1 bonus/4 levels to combat maneuvers. Looks great on paper, right? Not so much in play. When compared to the terror of a straight Fighter, their best comparison as what is supposed to be a front line melee combatant, they were terrible. This was due to loss of advanced weapon spec, loss of base rate of attacks, base thac0, you basically had to have elven chain, a huge difference in HP, and a deep level of MAD (causing all the effects that MAD does). The Bladesinger could not outdo the Fighter. Compared to the Wizard or Cleric (?). Uh no. Because Wizard or Cleric. What's more, the class didn't have any real glue to bring it all together to make it a coherent whole as a unit in play. They had some decent thematic stuff that was mostly just Jedi knock-off for elves. And no support to be able to actually cast spells and fight in melee at the same time.
Fast forward 18 and 20 years to the FR campaign and Neverwinter campaign settings and the releases of the 4e Swordmage and Bladesinger. Here we have a classes who don't suffer from any of the above problems. They don't suffer from multiclassing rules generally. They don't suffer from MAD. They don't suffer from lack of focus on "what they're trying to get done." The Swordmage is sturdy as all get out with the base chassis of a Defender. Through spell, steel, light armor, and magical warding, it controls the battlefield with instant teleports and reprisal for enemies attacking allies, and unleashes blasts/bursts of elemental energy. In cool Jedi fashion, it can fling its sword and pull its sword back to it. The Bladesinger has the base chassis of a Striker. True to the source material, it gets a bonus when spellcasting, but this bonus is no OAs when casting a spell. Also true to the source material, it relies on one handed melee weapons. It relies on light armor and an armor bonus due to its swordplay providing a shield bonus (like the SM's Warding but martial and not magic). While it can cast (powerful) Wizard Utility Spells (such as Shield, D-Door, Fireshield, Stoneskin, etc) and encounter spells (such as Burning Hands) as dailies, its defining features actually synthesize the idea of sword and spell. Its Bladespells allow for at-will channeling of small spells through melee attacks. It flurries with a melee attack (unleashing another Bladespell) when it uses its Wizard encounters. But its primary feature is the Bladesong encounter power which is a ridiculously powerful buff that grants bonuses to all defenses, to hit, a huge damage bonus, and an interrupting counterattack if struck in melee. The trick of the class is to extend Bladesong (so you're always, or virtually always, in the trance) and deploy what amounts to a sustained nova while controlling the battlefield (positioning and who gets attacked) with your spells and your OA reprisal if struck in melee. And neither are even remotely OP as they work seemlessly with 4e architecture.
2 different takes on the same shtick. A robust action economy ensures no AE issues. No prohibitive multiclassing issues. No crippling MAD issues. And distinctive suites of abilities and synthesizing glue that emboldens archetype (while even having diversity within archetype) and allows for front line melee combatants that has competitive parity with their peers.
5e would do well if they looked right there (and at the Avenger and Warden) for their support of the F/M-U archetypes. If they ignore those design requirements, its going to be Monk-like "looks great on paper, crap in game" all over again.
