I disagree with the other voices here. Although it would seem unethical to suggest that a professor has reviewed a document which has only had a student overview, there are two points to make:
- Peer review is about guarantees of quality, not about processes achieved
- The graduate's review may not be the final review submitted by the professor
In any review-based system, no one can guarantee that any review is perfect; only the aggregate effect of many such reviews creating a resource of high quality. Thus the review process for a highly respected journal is much more stringent than that of an 'easier' one.
On the side of the reviewer, this is a responsibility-based role, not a procedural one; what is important is that the professor is willing to stake a measure of their reputation on the review; if the graduate knows more detail of the field and the professor trusts their judgement, it could be a more thorough review than the professor could provide. As long as the professor judges the reviewer to be competent, that is their responsibility.
Finally, having a student review a paper may well be only part of the process; give it to a couple of students, see if they come up with anything. If they don't understand it, or they discover flaws, then the professor can review the paper herself in that light. Just because the graduate student has reviewed a paper doesn't mean that the professor will just pass the review on without comment.
The key, though, is that the journal is only asking that the professor provides responsible and competent feedback, and signs off on the review. How the professor reaches that goal is up to them, and peers that provide poor reviews should be excluded from the review pool (and associated privileges) by the journal. There is no magic to a professor reading a paper, and their judged competence to do so is only based on the university's assignment of professorship and the academic's history of research. Everything builds on expectations and responsibilities, not on some procedure having been carried out; the procedure is only there to make those responsibilities explicit.