fix(convert): decode JIS X 0212 0x2237 to a single character

[youdie006]

EUCJPToUTF8() and JISToUTF8() always write at least two bytes for a JIS X 0212 table value. 0x2237 is the one entry that maps to ASCII (~), so EUC-JP 8F A2 B7 decodes to a NUL followed by ~:

Encoding.convert([0x8f, 0xa2, 0xb7], { to: 'UNICODE', from: 'EUCJP' }) // [0, 126]

The ISO-2022-JP form (ESC $ ( D + 22 37) does the same. glibc iconv -t EUC-JP writes ～ (as in 10時～12時) as these bytes, so the NUL shows up in real text. Python's euc_jp decodes them to ~, and glibc iconv and WHATWG TextDecoder to ～; all return one character.

This writes values below 0x80 as one byte. Converting every EUC-JP, JIS and SJIS code to each target, only the six outputs for 0x2237 to UTF8, UNICODE and UTF16 change. Added a test; npm test and npm run lint pass.

Written with AI assistance (Claude); I have reviewed the change.

[polygonplanet]

@youdie006
Thank you for the detailed PR!
I confirmed that an extra NUL (0x00) was inserted in both JIS (ISO-2022-JP) and EUC-JP using the example code you provided. I also appreciate that the fix is minimal and focused.
This looks good to me, so I will go ahead and merge this. Thank you!