No doubt, it really is a thorny subject! 😅
"Immersion" can be a dangerous word, especially in the TTRPG context. I often see it invoked as an excuse to shut down conversation at the table, or to keep certain rules deliberately vague, which tends to just breed frustration.
In videogames, though, what people usually mean by "immersion" is the sense of being present in a world and/or an event. Plenty of factors feed into it, but it stays deeply subjective, and that's probably why it ends up feeling like a synonym for "everything I happen to like." For me, less reliance on extradiegetic cues and less visible designer interference is enough to feel immersed. For you, it might take some other factor entirely, or none of the ones I just listed. For someone else, it could be something completely different. That's the magic of it. Even so, the more recent literature has been making a real effort to pin down the factors these experiences have in common.
There's already an ongoing discussion around an "ideal level of guidance." Too much, and players get bored because everything is obvious; too little, and they get frustrated because they're lost. Personally, I think the sweet spot is a balanced use of diegetic and non-diegetic guidance. But narrowing the research focus, for now, to diegetic means lets us look at something useful in both scenarios: games that want to rely on diegesis alone to orient the player, and games that lean on markers, minimaps, and the like. In that second group, the two actually can reinforce each other.
To me, immersion sits right alongside "realism" and "living, breathing world" as terms that get thrown around pretty loosely. I often catch people conflating immersion with flow, for instance. But, nonetheless, it still is a very interesting subject to delve into!