I have come around to the position (argued by @adamsmith in a previous discussion on this topic) that a book review item type would be a suboptimal solution to the problem, since it would preclude me from noting the form in which the review appears.
Generally enter book reviews as Journal Article and include the review DOI. Those won’t be marked as duplicates.
I think I've said before that that should improve, but if the only case for a new item type is to prevent duplicates*, I don't think that's a convincing case. That just means duplicate detection needs to get fixed.
Review Articles are just another type of journal article, like editorials or research articles. They are cited the same. That isn’t going to change.