Its not about the books being too confusing and difficult for even grand masters. Its the fact they are simply not as practical ad doing the exercises on the same medium and platform you will be competing on. And it makes no sense to analyze other peoples games over your own. And it makes no sense to learn theory when you still fail at basic chess principles, tactics and strategy.
OK so that's where you're right and EVERY CHESS COACH ON THE PLANET is wrong. That seems perfectly legit to me.
You study other people's games, mostly GM games so you can build your thought processes, generally two epxert players, and see the minor things that are going right or wrong and how they evolve a position - that is in addition to spotting the inevitable blunders that most people games (sub 1800 or so) have.
If you can't see the difference between spotting your own blunders, what led to them, and how you might avoid them and AS AN ADDITIONAL part of your practice looking at why Fischer gaining space on the queenside, or Alekhine opening the c files at the right time created winning chances, then I would suggest you stick to playing League of Legends, or Grand Theft My Little Pony.
Pretty sure learning types or styles are mostly a myth; most people benefit from active learning (reading the book or watching the video WHILE taking notes or trying moves on digital/physical/mental board) instead of passive learning (just watching video, just reading the book from front to back, just listning).
Also pretty sure that chess masters don't have that much better generel memory than average people, just mouch better chess chunking.
I agree with your first paragraph which can rougly be summarized, by simply doing practice excercises on the board, playing games and analyzing your own games.
As for saying that chess masters don't have good memory you must be joking. They have exceptional memory which along with having good spatial visualization is a main requirement of being a good chess player.
I remember there was a study done by someone a while back in the 1900s. He got a bunch of masters and amateurs and set up the position and had them look for two minutes. Then, he’d give them an empty board and have them reconstruct them.
On positions from actual games, masters did remarkably better. But when they were completely scrambled resembling nothing like a chess game, the masters did not do much better than amateurs, which suggests that they chunk the board differently from amateur players.
Adrian DeGroot c.1953 off the top of my head. I could look it up, but I think that's the study you are referencing. DeGroot's work built on a foundation set by Alfred Binet, and became a foundation for the seminal work of Chase and Simon in the early 1960s.