Section 4, Encodings, currently contains the statement: "User agents must not support any other encodings or labels."
I don't think this restriction is acceptable. Some user agents have already implemented support for other encodings (for example, the Google search engine supports some Indian font encodings), and there's no good reason for them to stop supporting them and make web pages using them unreadable.
I think this should be WONTFIX. We should only support UTF-8 and the set of legacy encodings needed to successfully render the current Web.
Supporting other encodings just adds incompatibility and attack surface by definition without improving Web compatibility.
As for Indian font encodings specifically, pages that rely on particular fonts are already broken in the sense that they don't work correctly across different client systems. We should get Indian sites to use UTF-8 instead of perpetuating uninteroperable font dependencies of the 1990s.
Also note that the font dependency use case is server by x-user-defined for legacy purposes.
Furthermore, experience says that leaving things open-ended makes people introduce security holes by exposing Web-unsuited encodings to the Web. Being an unfortunate person who tries to clean up messes like that, I have zero sympathy for the request made here.
If there's a case to be made to add an additional encoding it should be added to the standard and supported by everyone.