Update the GB18030 encoder to spec-compliantly handle old PUA code
points via a direct byte lookup table (spec step 5). Bake the 18
GB18030-2022 code point updates into indexes.json and remove the
now-unnecessary patching logic from the code generator. Drop the
redundant hardcoded switch in the decoder's range function, as the
range formula already produces correct values.
Import WPT tests for gb18030 decoder, gb18030 encoder, and gbk
encoder, and register the worker variant in TestConfig.ini.
When the encoder encounters an unencodable code point while in jis0208
state, the spec says to emit ESC ( B (0x1B 0x28 0x42) to switch to
ASCII mode before returning an error. The encoder was incorrectly
emitting ESC ( J (0x1B 0x28 0x4A) which selects Roman mode instead.
This caused form submission using ISO-2022-JP to produce incorrect
escape sequences when replacing unencodable characters with numeric
character references.
Also imports the WPT iso2022jp-encode-form-errors-stateful test.
