My understanding of the view transition pseudo-elements is that they are all pseudo-elements on the root element. This means they don't really make sense to be allowed after ::part(), despite that the ::part spec says that all pseudo-elements are allowed after ::part().
If we agree in #10786 that the rules on what is allowed after ::part() are a parse-time restriction, then we probably need to decide whether the view transition pseudo-elements should be allowed. My current inclination is that they probably shouldn't be since they could never match, but I don't have a strong opinion.