The parser’s data structure is a more traditional Abstract Syntax Node design. Roslyn’s tree design is, IMO, a good internal structure for use in a production-grade compiler meant to be coupled with a developer environment (IDE and other tools) because the complexity pays off in terms of developer experience. However, Bob set out (as far as I can tell from the book and his talks/comments on the subject) to provide a fairly beginner friendly, introduction to the topic of language implementation. Within his chosen scope the parser implementation in Crafting Interpreters is more than sufficient and clearly presents the concepts intended.
