I had my own experiences with MDPI both as a reviewer an as a special issue editor. I'd say they don't do anything I'd call "unethical", but they definitely follow a kind of business-like approach to the whole process, trying to observe the rulebook and the deadlines.
As a reviewer, you get an invitation along with other reviewers. If you respond too late, the link simply expires. I am not sure what happens if several people respond quickly, but probably it rarely happens in practice: it often takes longer than planned to get enough reviewers, and if there is no response within a reasonable time, another candidate will be invited.
As an editor, you can leave the work to inviting reviwers to the MDPI staff or do it yourself. In either case, they notify you when all the promised reviews are received (or the deadline has passed), and you can see them all in the editor interface. So I really doubt that any review that was actually written won't be shown there.
I'd say that MDPI has an extensive ever-growing list of "special issues", which are edited by guest editors (like me). It's an editor's job to decide what to do upon receiving reviews, so the authors' experience depend a lot on editor's attitude. For example, seeing a poorly written review, an editor might ask for another review or simply look at the final verdict (accept / reject) without much attention to review content.
From a purely user interface perspective, the editor doesn't really have any incentive to prefer some reviews and ignore others. There is a box with a short summary of all reviews (like "Reviewer 1: acccept; Reviewer 2: major revision; Reviewer 3: reject"), so I doubt any reviews are lost at this stage.
Naturally, if the deadline has passed and we only have two reviews out of three requested, the editor might decide to make a decision without waiting the for the third review (especially if two people have already proposed rejection, there is little merit in delaying the decision), or to invite another reviewer rather than waiting for a reply.