To me, whether a concept gets the full class or subclass or feat or background treatment is really about, "how broad an idea is this? Is there room for 5-10 mechanical and narratively unique iterations that still feel part of a narrative and mechanically cohesive whole?" If the answer is yes, then it can be a class.
Some concepts are so broad, like a Sentinel or a Sharpshooter, or so narratively malleable, like a Criminal or a Sailor, that they're better suited to other pillars of the game's mechanics (feats and backgrounds respectively). It would be a real shame to have to take a level in Sailor to be able to make your Fighter a competent Sailor, or worse, to have to multiclass deep enough into Rogue to reach the Swashbuckler subclass to get those sailor abilities (not in real life, just a hypothetical based on the subclass name). The concept works better as a background occupation that any character class might have had before they became an adventurer. Sharpshooter, on the otherhand, is a broad idea of an expert archer/dagger-thrower/gunslinger etc, and it could work with any number of classes and subclasses that have ranged capabilities. There was no need to create a Sharpshooter subclass of Fighter (as appeared in UA ahead of Xanathar's), because that suggests that while a Rogue or Ranger (etc) could be a great archer, only a Fighter could be an expert sharpshooter.
So when we're thinking about, "is there narrative space for this class," we need to know if the class can stand on its own two feet as a mechanically and narrative consistent whole without gatekeeping concepts that would naturally fit into any or all of the other character classes. It makes sense to make Psionics, as a broad concept, available via feat, as an alternative to multiclass dipping into a hypothetical Psion class. This is akin to taking the Ritual Caster feat - it opens the door into a new world of abilities. Of course, Psionics can't rely on an optional rule-set like feats or multiclassing, so the question is whether it works as a full class or as a series of subclasses in addition to that feat.
As a parallel, I'd like to present the Arcane Gish - one of those rare concepts like the Warlord and the Psion that are "missing" from the game's class list but compelling enough to be a class (as they have been in past editions).
The problem with Arcane Gish characters is not that it's not a compelling class concept or that there isn't room for a dozen or more iterations on it. The problem is that each idea of what that concept means is wildly different from mechanics and narrative (as Arcane+Martial is a wildly broad concept). This is why we have Eldritch Knights and Arcane Archers and Bladesingers and War Mages and Oath of the Watchers and Hexblade Patrons and College of Swords and College of Valour and Arcane Tricksters and Wild Souls and Way of the Four Elements and Battle Smiths and Artillerists and Armorers (not to mention the rejected UA Stone Sorcery origin which leaned into the Arcane Aegis concept of 4e Swordmages) - there's so many different ways to iterate on this concept, and to try to unite them into a single class fails to satisfy the narrative and mechanical "feel" the fans of each archetype is looking for. By housing these related archetypes in different classes, it gives the concept a life that otherwise is overly general and broad, and allows a fine-tuned idea to be presented.
The downside of this is that 75% of the time, you're not an Arcane Warrior, you're a Fighter/Wizard/Paladin/Warlock/Bard/Rogue/Barbarian/Monk/Artificer(/Sorcerer), and if the core mechanics of your class don't synergise right with the particular Arcane Gish concept you're going for, you're a bit out of luck.
When WotC made the Mystic, they essentially tried to make the Arcane Gish a single class, only they were doing Psionics, not Arcane Gishes. This is a kitchen sink, and it's akin to what the Wizard was in the earliest stages of the D&D Next playest: barely a thing of itself, just a name and a single mechanic that links vastly different concepts together. This is what class families were in 2e, and what Power sources were in 4e. It's not a real class. This is the sort of narrative whole that is needed for Psionics (they need to all be related to one another in some way), but it does not a class make. The Psionic Talent die is a great way, imho, for doing this for all Psionic concepts.
If we want a "Psion" class, however, we need to figure out something a lot more focused on a single psionic concept, but broad enough that you can churn out subclasses for it. And those subclasses can't be "one feels more martial, one feels more roguelike, one feels more magical, one feels more divine, etc." That leads us back into the issue with the Mystic.
So we need a singular class that people will agree is awesome, that has room for more than just, "Telepath, Telekinetic, Pyrokinetic, Ectoplasm summoner/shaper." And it can't include the concepts of 4e's Battlemind/3e's Metabolist necessarily, because that one is so martially focused (better as a Fighter or Monk subclass). It could. however, include the Ardent if such an Empath concept is less martially focused than it was in 4e (never understood why the empath had to hit people with a spear to mess with their emotions).
But is there room for 5-10 more subclasses? The other main classes have a whole host of concepts screaming for them. Can we crowd-source some concepts? And if so, is there an agreeable baseline for the Psion? If there isn't enough subclass ideas for the Psion or enough of an agreeable baseline for it, Psionics is better off split between the various classes.