By GM-enforced alignment I would certainly include what you say, but also am intending to capture play in which the GM more-or-less unilaterally tells player what the moral meaning of their action declarations for their PCs is.
There are curious borderline cases here, whose existence is I think an offshoot of trying to reconcile AD&D's alignment techniques with authentic play: the player declares action X for their PC, the GM responds that it is an evil act, and the player in turn responds that it is nevertheless the right and warranted thing to do and follows through.
I can think of two possible "logical" structures to the situation I've just described. As I'll explain, I think one is typically a more apt characterisation.
(1) The situation is one of what Walzer calls "dirty hands" - situations in which
a, even
the only, warranted choice is an evil one. Walzer's go-to example of this is what he calls "supreme emergency" in wartime, which morally permits departure from the rule against killing civilians (his example is British terror-bombing of Germany prior to the US entering the war, which brought defeat of Germany by morally permissible means within the scope of feasible action). Characteristic of such situations is that the warranted choice is still understood to carry its moral taint.
(2) The situation is one in which the labels of "good" and "evil" are more-or-less rejected, either outright, or at least in the way they are being applied in this situation. Nietzsche is a strident exemplar of this, but there are slightly milder variants: Bertrand Russell, for instance, would talk about women being permitted to do "wicked" things once they got the vote, and he meant "wicked" in some conventional sense which he himself (being a suffragist) rejected.
I think the RPG situation I've described is characteristically like (2) rather than (1): the players (and their characters) repudiate the "conventional" moral labels that are being managed and applied by the GM, as "spokesperson" for "the gameworld". And so retain the conviction that what they are doing actually is morally permissible (contra a dirty hands analysis) and is classified as
evil only within a framework that they have rejected (or, perhaps, transcended).
(2)-type situations are closely connected to notions of authenticity.
I've done a fair bit of RPGing involving (2)-type situations myself, in both RM and 4e D&D, but not as a way of coping with or "sublimating" AD&D alignment techniques: I've made it clear that the god's moral framework is just that, and left it an open question for the players to choose their PCs' orientation towards the gods. I see this as, structurally although not in terms of content, a little bit like the "beyond just sandboxing" phenomenon that I discussed with
@kenada a bit upthread.