Reducto-ad-absurdum is a proof only when it takes into account everything in the argument. In life-or-death examples, other normally-included portions are dropped, making the argument something else entirely. It's just a straw-man attempt, not a valid point.
I used a different situation, it is true, but only because I have found being on the instructional end of cheating feels very different than you might imagine, so I guessed my parent was not familiar with it.
As a student who has not cheated, but has had to sit through countless lectures about cheating, and a few where large portions of the class did, I ask you and (other?) teachers this question: to whom are you teaching? The ones who are learning, or the ones who are cheating?
I understand that you're stuck with both, but either find motivation in the successes or consider another line of work - it's part of the job. Otherwise you're screwing the ones who did the right thing.
I am all for using analogies to reassess my views, but the degree of comparison should be similar for the analogy to work. I can't even process for a second how cheating is anything like threatening to kill someone.
Would you act friendly, trusting, kind and natural around everyone because up to 75% of them do not deserve your fear and suspicion?
It is perhaps a bit of an extreme thought experiment, but I think it illustrates handily how the human mind reacts in this kind of situation.