The suggested edits queue already requires a reason when rejecting (I usually choose
custom: this was an audit).
The purpose of the audit is to detect reviewers with bad criteria (and robo-reviewers... but I think that is not the primary goal) as Shog9♦ commented in Using the "flag" loophole to pass every review audit
The basic assumption is that most reviewers want to help, but some of them are just... really bad at it. There's also the danger of fatigue causing mistakes. Actively malicious reviewers are another story entirely.
Then, if you read the post and decided that it should be rejected, you have already passed the audit, and you don't need to waste time deciding why it has to be rejected (which may take some time in case of posts that could apply to several options)
I'm not sure if there are audits for "false negative" rejects (i.e audits where the expected action is accepting). There is a comment in the question I linked saying that only the close queue has them, but I don't remember any.
BTW, I think that the current system is good for those trying to do a good job when reviewing. When I fail an audit (and I must confess, it happens more frequently than I would desire) I understand that as a signal for taking some break before more reviewing.