Update integration with Encoding Standard

This changes a number of things:

* Corrects casing of encodings to align with the Encoding Standard.
* Cross-references all mentioned encodings.
* Normative: When encoding we no longer replace UTF-16 by UTF-8, but use “get an output encoding”
  instead, which does the correct thing for “replacement” and makes this kind of handling
  centralized.
* Removes the changing of UTF-16 to UTF-8 for URLs since the URL Standard handles that now.
* Normative: Centralizes encoding lookup for forms. This also fixes text/plain in the process, where
  the previous wording didn’t make sense.
* Normative: Removes the bogus “best of the set” algorithm for accept-charset, which no browser
  implements.
* Fixes the placement of "?" from the previous commit.