(1) Your group number 2 describes a sizeable portion of law school graduates who actually do end up in six-figure corporate jobs, as well as a decent (though smaller) portion of medical school students. Unless, only those persons who obtain the six-figure jobs out of graduate school are academically gifted. This would mean, however, that sizeable portions of the graduating classes of many such schools (even just the elite ones) harbor non-academically-gifted students. That's not an indefensible position. However, it would be rather odd and not terribly convincing on its face, as presumably only the academically gifted could have arrived there in the first place.
(2) Even assuming your argument goes through successfully, it is irrelevant to whether the person committed the sunk cost fallacy, which is the issue under discussion. Now, you could argue that subjective assessments of expected value always track the true underlying expected value (i.e. the actual expected value for any individual's objective abilities). In that case, all non-academically-gifted students who chose further education would likely be committing the sunk cost fallacy. That's not the case though, and you know that. Or you should, because it is prima facie absurd.
(3) It is relatively uncontroversial that there is a substantive disconnect between the supply of graduates and the number of jobs paying a salary justifiably commensurate with the debt-levels we are seeing. This has a couple of knock-on effects. For example, an across the board devaluing of degrees, which unjustifiably increases hiring requirements for lucrative professional jobs, further exacerbating prestige whoring, etc. This turns high school seniors deciding to attend elite private schools into even greater gamblers as the cost of an initial degree rises. Maybe they are or maybe they aren't rational to take that gamble (Are gambles by definition irrational? If so, pick a different term.). However, after having taken that gamble and come up short, they don't automatically become irrational in taking it again. Or, at least, the irrationality wouldn't necessarily have anything to do with the sunk cost fallacy.