Yeah, we've been over this before, we're never gonna agree. You want seven thousand different classes for every minute variation, I want few broad and customisable classes.
Alright. What are the pros and cons of each approach?
Your few-but-broad-and-customizable path has as its main pro that it's
simple (but there's a catch--more on that later). It keeps things constrained. Other benefits are a reduction of potential multiclass cheese, since you can't MC with a class you already have; efficiency of page space, and in general an economy of rules text (the amount of text required for a single class is usually far more than for multiple subclasses); and in particularly good implementations, a sort of common "language" of design.
What are the cons? Well, for one, as we've seen with 5e itself, classes tend to get flattened pretty dang hard because they can't pre-invest too much, otherwise they'd be stepping on subclass toes. This means classes tend to get stuck with really generic features, reducing the distinctive flavor of each class in most cases (Warlocks being a major exception): Action Surge, Arcane Recovery, Sneak Attack, etc. Conversely, subclasses are
really damn small. As in, rarely more than 4-5 features in absolute sum, all of which must be small enough and light enough to fit within the chassis established by the overall class.
And that's where the "it's simple" has its catch: for it to really be customizable with meaningful depth, you have to smuggle back in the complexity
somewhere. Otherwise, as
@Minigiant just put it, you end up having to strip out the elements that make the things meaningfully distinct.
The main con that everyone and their brother brings up with having multiple classes, which you mocked with overweening hyperbole, is that having more basal classes means more stuff to learn. DMs have to learn more, players will feel they need to learn more in order to make a meaningful choice, designers need to factor in the combinatorial explosion of
à la carte multiclassing, etc.
But if
slightly more complexity is the price paid for having classes that are actually weighty, that actually
support the flavor and gameplay-experience they're billed as offering, then I'm absolutely willing to pay it. And that's why I keep saying things like "an Eldritch Knight isn't a swordmage. It's a Fighter who moonlights as a weak Wizard." It doesn't deliver on the concept; at best, it weakly imitates it. And that's why we
keep getting more "it's a spellcaster...who uses swords!" and "it's a fighter...who uses spells!" subclasses--because the fundamental, underlying desire is
not satisfied by these weaksauce, inadequate echoes of the class concept folks want to see come to life.
And, to your "seven thousand different classes" hyperbole: I've already gone through and pared things down to a realistic number. I think D&D has in it somewhere between 18 and 24 distinct classes--generally falling closer to the middle of that range unless you get real particular, or start adding more obscure options like "monster" classes or similar. Not even double the number of classes present in 5e as it currently exists, at the absolute most--and possibly not even 50% more.