I think @hawkeyefan's post about Wilds Beyond Witchlight is a better answer to this question than I can give!But is this a problem?
I'm just replying to reiterate that the OP isn't trying to identify a problem, just a phenomenon.
I see a connection between these statements and @Manbearcat's "First Rule of Planchette Club". Some RPGs seem a bit hesitant to come out and state their premises/rules. For instance, the game text might say that your job as a player is to portray your PC, but then the game play that is presented depends upon not conflicting with other PCs, or not splitting the party, or following the GM-presented premise, even when that means not faithfully portraying your PC.The time to reject the premise is when the game is proposed, not once you've started in on the game. Rejecting the premise once you've agreed to play is just being a jerk.
Again, while this is metagaming in the sense that it's outside of the rules as laid down by the game, it's not something that should actually be coming up regularly in play if players are accepting the premise of the game they're playing.
The "problem" in these cases is unclear/confused rules text.
I think players need to know the ground expectations of the game they're playing in order to engage with it appropriately.
If you don't tell the players that's the kind of game you are running they are probably going to be lost with how to engage with the game (unless that's the type of game they usually play).
What I'm about to post comes out of a fair bit of thinking about the topic, but is nevertheless new in my attempt to actually write it down, and so at this stage remains fairly conjectural.I don't know if I agree with you that because the players are aware that there's a structure to play that means that you can't eliminate or at least reduce the meta aspect of the way they engage with that structure.
I think that most RPGing needs to focus on non-typical facets of life. Even the Pendragon/Prince Valiant example, which takes knights errants who by definition live quest- and joust-filled lives as its protagonists, needs to gloss over the polishing of armour, tedious dealings with squires, weeks of riding through the enchanted forest before arriving at the strange castle, etc. This requires at least one participant to make editing-type decisions about the fiction - what aspects of setting, what aspects of situation, what aspects of protagonist decision-making (motivation and/or action), are foregrounded?
Different sorts of game structures, allocations of responsibility, etc do this differently.
In the "adventure" model, the GM foregrounds NPCs who (deliberately, by approaching the PCs, or inadvertently, by dropping all their files in front of the PCs) bring their plights to the attention of the protagonists. The players then make decisions about the protagonists based on metagame/OOC knowledge (or recognition) that this is what the GM is doing. @hawkeyefan's post shows this at work (and also not working, because the players didn't recognise that the GM was engaged in this process).
In the Traveller model, the players foreground the "freelancer" motivations and consequent actions of the protagonists (by spending their week at the Travellers' Aid Society so as to generate a patron encounter check). This sets limits on the viable range of protagonists (no truly shrinking violets, for instance) but seems to require less player decision-making based on metagame/OOC knowledge.
In the knight errant model, the GM foregrounds the moments of excitement ("after weeks of riding, you come to a castle"; "you enter a forest glade and see a sinister-looking knight astride a jet-black destrier"; etc) but the players don't really need to act on OOC knowledge at all. The castle's occupant, or the sinister knight, may be related to their over-arching quest (eg be an agent of Morgan Le Fay) or may not be, but the players don't need to form beliefs about that OOC one way or another in order to make decisions about the protagonists.
The Apocalypse World model seems a bit like the knight errant model except that it is based less on vignettes established via GM editing of time and place, and more on situations that engage protagonist motivations, based on GM editing of non-protagonists characterisation (ie NPCs - who I think have to be more vibrant and nuanced than the cardboard cutouts who can populate the worlds of knights errant) and GM editing of how situations unfold (of course, in AW, structured by the results of checks).
In both the knight errant and AW model, of course the players have to accept the basic premise of play - we're knights errant or Here we all are trying to make a living in Dremmer's hardhold - but I don't think there is much need for them to make out-of-character decisions in order to make play happen.