The interviewee is not your friend, he/she do not know your brand of humor or internal jokes. Additionally, he/she may have cultural differences.
You talked with someone an hour in an unbalanced stressful job related situation and you really think you know something about their personality? Behavioural researchers think that you can't know much. I wouldn't be so judgmental of people you've just met. You should be more aware of the limitations of the interview platform and stick to what you can deduce. (Which is not, that a person is toxic, because he/she felt laughed at)
This is the problem. There are many situations in a career which aren't "normal" and will make people nervous. Giving presentations. Talking to the CEO. Or the boss's boss. Or the leading expert at X. Being in a meeting with all the execs. Or in a performance review. Meeting with clients/customers/potential customers. If every time a person is nervous they assume ill intent of others at the slightest hint, they are going to be problematic. They don't intend to be problematic, but they are.
The point of a mechanism like that given is that you don't have to make a judgement. It's self selecting. In this specific scenario, the kotlin example interviewer a) laughed b) offended a person and c) probably wasn't even aware they had caused offense and d) kept the person out of the company.
If b) had gone down the other path, the person isn't offended and (all else equal) is still enthused to work at the company.
Edit since I can't reply to your comment: if you lack the emphaty to understand why people may be very reasonably insulted in such cases, you really shouldn't be conducting interviews, IMHO. The irony part is that you actually describe a behaviour that "assumes the worst of others", as you said.