From the counter argument:
> as a matter of fact, there is something really interesting about a mouse pointer feeling less like a deity floating above it all, and more like a regular in-game actor.
My counter-counter argument would be a general principle for UX designers: Are you designing a game or a tool? If you're designing a tool, don't put cutscenes in your software.
I think games are special, because their explicit purpose is to deliver an experience and often also tell a story. Within that context, I'm fine with having control restricted or yanked away if it's in service of something meaningful in the game.
The same is not true for tools (even in-game tools actually), where I want to complete some kind of task in the most efficient way possible - and often only I know the context of that task.
Unfortunately, that stuff has already seeped into UX design in a lot of forms, in particular as random "new feature" popups that usually appear at the worst possible moment and cannot be shown again. In situations like this, I'd value predictability much more than the coolness factor of a game-like UI.
Even in the context of game development, if you don't provide a "skip cutscene" interaction, some players get very upset. The desire to tell a story can be at odds with the player's desire for control, even there.
> often only I know the context of that task
This doesn’t account for cases where the system is semi-automatic, such as a tiling window manager (you don’t set all the dimensions manually). If the automatic part is to stay at all, it should make certain assumptions for the user.