The consumer of the rating (in this case, the code author) usually needs to know one thing above all else: Do I take action or not. A two-point system is the default.
IMO, in a code review, the feedback should be either expected action or not. "I recommend you change this" or "FYI, consider this", aka, "required" or "optional".
Ultimately, as the author, I want to know if I'm being told "please change this" or "just FYI". Details can be haggled if necessary.
The strength of the recommendation isn't relevant unless there's push back, in which case the details can be haggled for the situation at hand. But trying to do that bucketing up front sounds like extra work that's usually unnecessary. As long as the reviewer is pursuing productivity, they can adjust their recommendation as they learn more.