A timeline is just a best guess about how everything is going to work out, based on known factors. They're subject to change.
I'm talking about the timeline before it has been changed.
In my GM prep scenarios upthread, in scenario 2 the GM had prepared a timeline, which s/he draws upon in order to describe the Garden Gate to the players, depending upon the ingame day that their PCs arrive at it.
My point is that reading out the description from the appropriate day of the timeline is just as much GM authorship as preparing a single freeze-frame description and reading out that.
Of course no decent GM will stick to a freeze-frame description if other action declaration has declared it irrelevant. For instance, if the PCs fill the dungeon with Cloudkill then when they get to the torture chamber the GM won't be reading out the freeze-frame about the torturer taking the heated brands from the oven. S/he will be describing the torturer and the prisoner both dead, perhaps embellishing with some description of what the torturer was up to when the gas came into the room and killed them both.
It would have been less problematic, in that it never would have happened. They need that mandatory encounter in order to tell the story that they were trying to tell, where this corpse is the only one with the password to get through the magical lock.
I don't really understand what you mean by a "mandatory encounter". There is no law that obliged your GM to run this encounter, and no Paizo operative even figuratively, let alone literally, holding a gun to your GM's head.
Your GM chose to run this encounter, presumably because s/he (?) thought it added something to the game.
In your preferred style, suppose the PCs are trying to infiltrate an ancient tomb, and only one person knew the magical password, and that person has died: then it would seem similar to the scenario you are describing. And if that person's corpse were to be eaten by a demon (which is significant, I assume, because it precludes Speak with Dead), then the PCs would be stumped - just as in the scenario your GM is running.
Presumably, furthermore, as per your hypothetical timetable above, a sole PC could stumble upon the corpse-eating demon and realise that, if s/he does not stop the demon, the party will have no chance to recover the password and thereby break into the ancient tomb.
You would then face the same choice as your GM faced: what do you, as GM, do about the fact that it seems the players have no chance (within the mechanical parameters of the game) to pursue the path they want to pursue.
From what you've said about your style, upthread, you would have the demon eat the corpse, either killing or ignoring the sole PC, and hence the tomb-infiltration adventure would come to a peremptory end.
Your GM, from what you've reported,
didn't want the adventure to come to a peremptory end, and so used illusionistic manipulation of backstory to create a new pathway for the PCs.
The difference between your style and what your GM did, as far as I can see, has nothing to do with the encounter itself, but rather to do with how you integrate the consequences of the episode into the ongoing fiction of the campaign.
I'm saying that it takes more "force" for the GM to cause an unlikely event than a likely one. If you walk down a street and the GM says that nothing happens, then that requires less GM force than if you walk down the street and the GM says that you find a bag of money on the side of the road.
Another way of looking at it, you could say that the GM has a finite amount of 'plausibility' that can be expended before players lose interest and/or call shenanigans. The amount of plausibility expended with an event is directly related to its perceived probability.
This is a completely idiosyncratic definition of "force". It implies that naturalistic or pedestrian descriptions write themselves - which obviously they don't!
Using the notion of "force" in its ordinary sense - of referring to unilateral specification by the GM of the content of the shared fiction - the GM can use force and railroading to produce a naturalistic or pedestrian game just as much as a gonzo one. For instance, every time the players declare high-risk or over-the-top actions for their PCs the GM declares failure, without reference to detailed mechanics or dice rolls.
That would be a game with a very naturalistic fiction, but high in GM force. And it is actually a fairy common type of railroading/GM-blocking that one reads complaints about on these forums.
I would counter that they want agency WITHIN THE WORLD.
If the GM reads from boxed text "You hear X, and therefore decide to do Y, and arrive at place P, where you observe Q and therefore choose to do R, which has consequence A, etc" then the PCs are exercising agency within the world.
But the players need not have bothered to turn up!
In other words, I don't see how you can say anything meaningful about player agency if all you are talking about is actions taken by the PCs within the fiction.