I think browsers should render any content type in a way that provides functionality for a user; for example, I think it would be great if when loading a JSON file, I could toggle between the raw data provided and a pretty-printed, syntax-highlighted, collapsable tree view of the JSON file. I think it would be great if when loading an MD file, I could toggle between the raw markdown and a rendered version. Certainly when I load an image or audio file outside a webpage, my browser decodes renders the content usefully.
Given these cases, why would XML files be any different?