Despite the shared system, there is an issue with simply balancing a 5e videogame with the 5e Tabletop CR system. This is because every PnP game is by default a "no-reload, ironman campaign".
When you adapt that into the videogame world, where reloads are expected, that is a massive layer of difficulty completely removed.
Most PnP encounters are balanced in a way that your chance of losing as a party (TPKing) is actually extremely low. "Deadly" encounters are scary in the context of a "you only live once, no reload mode", because there's a chance of an entire campaign ending. In a videogame, that's closer to the baseline requirement for challenge. BG3 needs to be balanced against the playstyle of videogames (i.e. with reloads), not ironman mode.
Per the encounter descriptions in the DMG (pg 81):
Easy: An easy encounter doesn't tax the characters' resources or put them in serious peril. They might lose a few hit points, but victory is pretty much guaranteed.
Medium: A medium encounter usually has one or two scary moments for the players, but the characters should emerge victorious with no casualties. One or more of them might need to use healing resources.
Hard: A hard encounter can go badly for the adventurers. Weaker characters might get taken out of the fight, and there's a slim chance where one or more character may die.
Deadly: A deadly encounter could be lethal for one or more character players. Survival often requires good tactics and quick thinking, and the party risks defeat.
In the context of a videogame, anything below hard can be considered a "Trash Mob". Encounters of this difficulty is bad in a RTwP game, but at least there, you can turn on the AI and just watch it play out quickly in real-time. It's basically unacceptable in a slower, turn-based combat game with no "auto-resolve".
Even "hard" encounters are designed for parties to never lose - at most, those drain your resources (which is why they should be in the game, as they do offer a strategic decision). If you factor in the extra loot and infinite resting in BG3, it's clear why Larian needs to use more "deadly" encounters* under 5e to ensure there is a degree of actual challenge sprinkled throughout.
*As long as it's "deadly" within a reason - i.e. the Minotaur encounter is a 2100 xp encounter vs. the 2000 xp threshold of a 4 people level 4 party, which is still fair.
No-reload, ironman settings do not increase the difficulty of the game, they increase the cost of failing. The ability to reload or to save scum your way through the game has zero impact on the actual difficulty of an encounter, it just means you can inevitably make your way through it if you spend enough time replaying it. And ultimately it means that you always have the potential to finish the story regardless of the difficulty given enough time. So I completely disagree on your video game adaption theory.
I agree that most pnp adventures are balanced around challenges appropriate to the level of the adventurers, but you must not play much if you think that player deaths are uncommon even on a medium difficulty encounter. That is why spare the dying is a cantrip and revivify is a 3rd level spell. I've recently DM Lost Mines and had a Red Brand thug (1/2cr) take down 1 2nd level PC and put another down to 4 hp at the top of the first round of combat because of lucky crit on the first attack and a high damage roll on the second attack. In that encounter there are 4 1/2cr thugs making it a medium encounter and it ended with 2 players down and the other 2 severely injured. This is not uncommon, particularly at lower levels. In fact I'm pretty sure that you noticed that big text box underneath the descriptions you quoted talking specifically about this issue particularly at lower levels, and that how some monsters challenge rating doesn't reflect that some of their abilities can make it almost impossible for lower level characters to overcome. I love how the guide specifically points out that a CR 2 ogre can 1 shot a 1st level wizard, which goes right back to my point on how the minotaur fight works in terms of a numbers game.
In terms of the tactical model of the game, it's not difficult, it's just broken IMO. I think the increased level cap will deal with a lot of these issues, because the players will have the tools and hp to deal with the encounters. That won't fix the AI, and the fight cheese strategy of using height, line of sight and hide to kill your enemies, or infinite resting, or barrelmancy, etc.. but maybe it will make it a bit less save scummy.