feat: support iso-8859-8-i and iso-8859-8-e charset labels

[baptistejamin]

Hey there.

iconv-lite is used by a library we rely on: mailparser, which we use at Crisp.

We've received Hebrew emails with charset="iso-8859-8-i", and they aren't decoded properly by iconv-lite.

Right now, encodingExists('iso-8859-8-i') returns false and decode(buf, 'iso-8859-8-i') throws Encoding not recognized.

iso-8859-8-i and iso-8859-8-e are valid ISO charset names. The -i / -e suffix only describes text directionality (logical vs. visual BiDi order).

What this PR does

• Adds iso88598i and iso88598e as aliases of iso88598 in encodings/sbcs-data.js, next to the existing hebrew / csisolatinhebrew aliases.
• The canonicalizer in lib/index.js already strips the dashes (iso-8859-8-i → iso88598i), so both the dashed header form and the canonical form now resolve correctly.
• Regenerates types/encodings.d.ts
• Updated tests

The code was partially written with Cursor and Opus 4.8, and I reviewed it myself; it's an easy and safe PR.