I believe the difference lies in the use of far more discrete sounds and a lack of a visual element.
The TV/projector screen rivets your attention. As much as we appreciate high quality sound and the use of surrounds, we pay less attention to the sound than on what's happening on screen, plus what's happening with the story, the characters, etc. (well, if it's a good story/character!) The use of surrounds complements this, as it's usually unobtrusive ambiance with some discrete sounds added as well. And even if there's a lot of discrete stuff happening in the surrounds, it comes as a result of what's happening on screen, and being elevated, doesn't distract from what's happening on screen. The surrounds don't necessarily blend with what's coming from the front three speakers, so it's not necessary for the surrounds to be at ear level, and again, needs to be elevated in order to let that ambiance fill the room rather than be focused on the listener.
But with m/c music, there's far, far more going on in the surrounds -- at least with the music I listen to (some of which were mentioned in earlier posts in this thread). There's generally far less ambiance, and more discrete sounds -- sounds/instruments that are intended to blend with the music from the front three speakers, more so than with movies/TV. That blending makes it more important for the surrounds to be at ear level. (Which is why for movies/TV, having surrounds that aren't timbre-matched to the fronts is less important; with m/c music, the recommendation is to have five identical speakers all around.) Plus, not having a visual element (the screen) to anchor your attention, having elevated surrounds creates an imbalance that is very noticeable, and not at all enjoyable. At least for me.
My little summary: If the surrounds are meant primarily for ambiance, then yes, they should be elevated so that they're unobtrusive and the sounds will be more diffuse. If the surrounds are meant to be more discrete and to blend with the rest of the sound, then ear level makes more sense.
At least, that's my take.
cheers,
supervij