EPT - at least in its classic OD&D-ish version - does not strike me as simulationist very much at all. It's kind of like NWP-proficiency era AD&D, but its resolution processes don't model in-fiction causal processes, and thereby determine the fiction, very much at all.
I found this difference in intuitions interesting. I joined EPT as a player for several sessions. RQ was the second RPG I owned, and I GMd it for 3-4 years. Generally, I'm more engaged by what I intuitively think of as simulation of cultures (rather than combat). Thus what fascinated me about EPT was the simulation of the distinct cultures of Tekumel, and in RQ it was Glorantha. Crushes to chest didn't excite me nearly as much as the cults, the glowline, mountainous giants, dream dragons, rune lords and priests, true stone (the spike), Mistress Race trolls and so on.
On reflection, I feel that the above is probably better identified as immersionism rather than simulationism. Taking the latter to want realistic feeling, detailed mechanics and results (where real is grounded in our real world, and the details are those we expect from life).
Simulation is married to gamism in games (more broadly) that aim for the highest possible realism (e.g. Harpoon or Advanced Squad Leader.) Possibly the roots of simulationism are thoroughly entwined with gamism. Albeit as you noted gamism traditionally demands an even playing field and a determined commitment to simulationism would not (presumably, although that could be more an immersionist concern). In any case, I believe there is scope in future for more sophisticated players to set that requirement aside: asymmetrical games, and those of unequally distributed challenge, can be even more gamist.
Simulation needn't get in the way of a premise or dramatic character development. Albeit devising, grasping and upholding complex realistic rules is effortful, and that effort can drag focus onto the simulation over other dimensions. Any really crunchy game seems to risk that. Maybe BW gets around it through encouraging openess, so that participants work through the rules together?
It's the core of purist-for-system (or "process") simulation RPGing, which I think is pretty key to what the OP has in mind (though the OP isn't using that specific terminology).
Simulations are necessarily approximations. Every playable game must pick and choose its subjects. Where we have rules for hawk speed, we might lack rules for hawk reproductive behaviour. Do merlins and condors have the same speed and endurance? In all winds? I don't reall any game mentioned so far realistically treating recuperation and long term disability. Or the psychological trauma resulting from participation in massacres.
My point isn't to say those things must be included, only that if our data model represents fireballs in squares, that choice isn't simulationist in isolation. The test has a large subjective component: do players feel like a fireball exploded realistically around their characters. The rules need to specify the sensations as much as the area and burns. Should it be the same in an enclosed room versus an open field?
For probably a majority of game mechanics, we lack validating data sets to know whether the model and simulation is really predictive of how things would go for the given phenomena in our real world. The question is probably less - is reality reflected accurately? - and more - does the game
persuade us that it is reflecting reality accurately? That last might be accountable for ambiguities or conflations between immersionism and simulationism. Often detail is seen as realistic, even though adding fields or dimensions to a model is no guarantee of accuracy and can produce system instability (less granular estimations are often more likely to be right).
How would things go for the given phenomena in our real world? Is that what a simulationist wants that an immersionist doesn't care about? The latter wants to know how things go for the phenomena in the
imagined world. Firecubes could be "realistic"in some imagined world. The simulationist wants it grounded in our world: the one we as players live in. (But we don't live in Glorantha, or Tekumel!) On a multitude of details, both are bound to use the same reference point (our world), and both are perforce using approximations and incomplete models. So the test might be better put - does it go how I feel things would go, for some given phenomena that I choose to focus on, in my chosen reference world?
