"I think the problem with playing something like 2...Nf6 is that White can push d5 now and get back into Benoni territory and Black doesn't get the traditional pressure down the e-file since the pawn is not yet on e4."
That would be great fot black since the e pawn must later go on e4 to supposrt d5, i.e. white has wasted a full tempo. An illustration of why Nf6 is slightly superior to d5: it gives white more chances or doing something really silly.
You proposed plan is entirely sound with the "problem" that you can play it with a 3...Nf6 move order while cutting white options. c3 is not that frightening in your line (i must repeat that is far from clear if black is going to have time for a minority attack and he has not at all an automatically slightly better position) but would be even sillyer against 3...Nf6 since blak will refrain from taking and just play 4...e6. So in a certain sense i think your proposed setup proves my point that Nf6 is a somewhat better move. No surprise since it is by far the most played.
I think the problem with playing something like 2...Nf6 is that White can push d5 now and get back into Benoni territory and Black doesn't get the traditional pressure down the e-file since the pawn is not yet on e4.
I definitely like setting up with Nf6, g6 + Bg7. So something like 3...Nf6 4.Nf3 g6 5.Bd3 Bg7 6.O-O O-O and White has to show his hand first, does he want to play c4 and leave d4 potentially weak, play c3 and give Black the minority attack option, or forego either plan for now and develop the knight to c3? Black doesn't have to commit to anything just yet, he can play sensible developing moves (I wouldn't rule out the possibility of the bishop coming to b7 or a6 either) and strike back with d5 at an opportune moment.
The pawn on e3 isn't that great in the Benoni though and white will waste a tempo moving it to e4 later on. Blunt that bishop on d3.
Nf6 is a good move to play and perhaps you can also play d5 later,g6,Bg7,etc in a sort of "Grunfeld" formation like you say.