You are bringing too much of your real-world common sense into this topic. Video games can do as they damn well please. I've played games where the profile of a jump is straight-line up, straight-line down, with no acceleration concept relevant at all, which also eliminates "terminal velocity" as a concern... it's just velocity, period.

Unfortunately, the same is true of the authors of this paper, who assume with no evidence, or perhaps rather against the evidence, that standard Newtonian formulas hold and therefore they can compute "accelerations" and such. Newtonian formulas do not hold in the Mario-verse, or in platformers in general.

(In some cases the real formulas may be modeled on Newtonian formulas, but the full set almost never holds, "equal-and-opposite force" in particular. Please note that citing a single counter-example is insufficient to disprove "almost never"; I can do that too, but it doesn't change the fact that most games use physics only loosely related to reality.)

I was shocked at the concept of acceleration for this very reason. I thought there was just velocity up and velocity down because I got my notions of physics from video games.

Yeah, terminal velocity is relevant over a short drop if the acceleration is high and the terminal velocity is low (both true in mario)

Looking at the table they made, most of the falls are about half a second long. I'm too lazy to go test this but a lifetime of experience playing mario games makes me feel like this is far more than enough time to reach terminal velocity

But Mario's terminal velocity might be much lower.