Despite the title and body asking "What is exactly 'This question doesn’t meet a Mathematics Stack Exchange guideline',” the post concludes with this:
I don't want a link to MSE guideline. I have an experience in MSE review that, MSE (AI) tests us sometimes. I want to know is this vote-to-close a bad result of that test? If not, why don't reviewers write down their reason to help to the OP how s/he can improve his/her question?
The Review Audits never result in casting closing, deleting, or reopening votes. They are simply "tests" as you say. The interaction of audits with attempts to leave Comments varies from causing failure of the audit to revealing that the item is in fact an audit. See Types where Comments are allowed (or not) in Review Audits for more discussion of that aspect.
The Review Queues are not especially convenient for reviewers to write customized Comments, though it is possible to do this with a little extra effort. The workflow guides reviewers to make decisions about the quality of items in the queues. The review is typically being done with a "keyhole" view of the post (either Question or Answer), and the best efforts often require stepping away from the immediate browser tab/window and searching for potential duplicates.
The number of comments left by a user (and not deleted) can be seen from their activity page on Math.SE under the All Actions tab, where one can also find the number of reviews done. While the fraction of comments left in the course of doing reviews is not easily discerned, it seems to me that comparing the number of comments to number of reviews is a reasonably good indication of how much effort was made to help users improve their posts.