I don’t really have especially coherent thoughts on this but my entirely unresearced theory is that since we are so good at finding meaning in things we convince ourselves of some interpretation based of off very little information. Like have you ever seen a kid making up a story for a book when they can’t read properly?
In early stages of learning just being able to follow a story is a big achievement (and rightly so) and it is easy to trick oneself into thinking that because you can follow the story that you understand most of what is being said. It probably isn’t true. I think we do a pretty good job of ignoring what we don’t understand and filling in the cracks without noticing.
In the more intermediate stages I would say the problem is one of not knowing what one needs to look out for. I think you need to either see the same thing a lot of times when immersing extensively or you need to really question everything very closely when doing intensive immersion (“Okay, they used が here but は is grammatically acceptable so why did they use が and not は?” etc). The answers might not always be that deep but it will lead to a more acute sense of what you are “whitenoising”.
Obviously comprehension is not something that is black and white which is what makes this problem particularly insidious. The two contexts in which make me realise what I am missing in terms of comprehension the most are when re-watching or re-reading and just during conversation. When re-watching or re-reading, normally after a few months, I will notice so many things that I didn’t catch the first time. It also makes me realise that I have improved (when I feel like my ability never changes) so it is a good exercise. On the side of conversation, you are really forced to consider if you understand something as you have to reply. If you don’t understand properly then your reply will be nonsense. If you are speaking with a friend they will probably rephrase what they said and you will realise your error. If you are speaking with a stranger then they will probably just nod politely and say 日本語上手 (I am only half-joking).
Another reason why this situation comes about is probably comes about is the fact that expression and expressive ability are a lot more complicated than the basic version found in grammar resources. Bunpro is one of the more fairly comprehensive resources but it wouldn’t be possible to include everything.
As you get better obviously this kind of thing happens less but it is just part of learning the language. The flipside is that if you don’t understand something now then that too will also become comprehensible after some time without you really noticing. That’s probably the nice version of this effect.
My final thought on this, which maybe people won’t agree with, is that if you compare your comprehension to your native language you will realise how bad it is even if you can follow everything. If you had to write a standard high school essay on something like “How does the author create tension in this scene?” or something then could you do it as well as you could in your native language? Some people might say this is unfair as a metric but if your goal is to have a high level of Japanese then being able to write a high school style essay is actually a low measure (depending on your educational background). My own big goal is to be C2+ so being quite strict about exact understanding makes sense but you can enjoy a lot of media without getting close to that level so at a certain point the question of precision when understanding things is just personal.
Sorry this was a brain dump - I have never put my thoughts together on this before. It is a good topic though.