One of the problems with the way that multi-classing has worked since 3e came out has been that it encourages optimizers to dip into classes for front-loaded benefits without the character making any meaningful commitment to the class.
If you're going to present this as fact as opposed to opinion, an you please provide some collaborating evidence. I've been very happy that for the most part the PHB classes don't give out major boosts until 3rd level or higher. There are some decent things you can get at 2nd, and Cleric domains are the obvious outlier, but my experience is that it isn't a problem in 5e.
I miss the old days of characters starting out with multiple classes, like the: Cleric/Magic-User, Fighter/Cleric, Magic-User/Thief, Fighter/Thief, etc.
While dreaming up character concepts for 5e I found myself yearning for the ability to create a Sacred Fist (Monk/Cleric) that wasn't so watered-down as to be useless. One of the things I like about 5e is that the Players' Handbook came with a solution to this baked into the system. By using the Eldritch Knight and Arcane Trickster as inspiration I have created numerous multi-class options that replace the standard sub-classes.
I really like having hand crafted subclasses like Eldritch Knight, Arcane Trickster, etc. They allow us to have a primary class and then touches of a second class hand crafted to that second class.
Though to me they aren't at all a replacement for the old Cleric/Magic-User or other combinations where you split your XP evenly and advanced in both classes. In those you'd often end up only a level or two behind and both classes really contributed to play and to your power. With the subclasses, it's one main class and the other flavoring it, but not near equally.
I am interested in hearing any feedback or constructive criticism. I did not include options for Clerics because the Domains are vital to a Cleric and give enough options to cover many options. I left out Paladins because they should be focused on fulfilling their Oaths. Warlocks are left out because they need a Patron to make sense, and Sorcerers because they need their Origin.
One problem is the number of combinations. In order to hand craft the combonations, there is little overlap. The Eldritch Knight and the Arcane Trickster both bring in some Wizard, but besides the common 1/3 casting neither handles it the same, giving it a different spin to meet the main class. Same if you compare Eldritch Knight to Blade Singer - very different classes, no design overlap in the subclasses. So what we'd be looking at is 12 classes each with 11 subclasses, so 132 subclasses just to cover "primary plus a bit of another class" multiclassing. It wouldn't cover variations in the classes themselves, like different types of sorcerers or clerics or whatever in that 132. And that still leaves out "I want the classes to be relatively even" like the cleric/magic-user combos that are you goal, and forbid the number of combos if you wanted a triple class like the classic Fighter/Magic-User/Thief.
The multiclass system is an elegant way to handle generic mixing and handles different amounts, with 5e doing a decent job or removing cherry-picking though giving things at later levels and moving back getting ASIs.
While I will like well crafted subclasses to hit a specific feel, I think it would be a lot of bloat to try and mimic a multiclass system using subclasses.