When talking about fiction-first games the fiction refers to the shared/established fictional situation the table has agreed upon. It does not include the conceptions that the GM and other players might have about the setting or their characters. Because those conceptions are not meant to be binding on play - things not revealed on screen are not meant to be binding or held on to by any participant. This is a fundamental part of the overall structure of play that needs to be understood.
Redefining the fiction to be inclusive of all our personal conceptions of the things we "own" is at odds with gameplay structures that are fundamentally grounded on the idea of a shared fiction.
I notice that Stonetop states that the setting book is intended for GM rather than players. It seems that there is going to be fiction that the group agrees has standing even outside of that which has so far entered their shared ongoing narrative.
However, Stonetop also defines
the fiction much as you have. Perhaps we should maintain the twin concepts of
the fiction and
setting (as I have proposed in this thread) to facilitate discussion about them and their relationship. To my observation the standing that
setting has does not appear to be all or nothing: it's
probably be true that The Crossroads is unsettling but that isn't fixed until it's said at the table (enters
the fiction). I suspect that's accurate to a greater or lesser degree of a wide range of modes of play (including some supported by rulesets characterising themselves as fiction-first.)
I really liked your discussion of
the fiction and am not trying to nickel-and-dime it. Rather I'm trying to understand how many or most TTRPGs really do have
setting that hasn't yet entered the fiction yet seems to have some sort of standing. That standing might be dismissed as not guaranteed, but for me that unhelpfully ignores the observable influence setting has on what enters the fiction at the table.