fix: strip and replace emoji that have a skin tone modifier

[spokodev]

Problem

strip() and replace() leave skin-tone-modified emoji in place, even though which() and unemojify() recognise them:

const emoji = require('node-emoji')

emoji.strip('X👍🏽Y')            // 'X👍🏽Y'  (expected 'XY')
emoji.replace('👍🏽', () => 'Z')  // '👍🏽'    (expected 'Z')

emoji.which('👍🏽')              // '+1'    (recognised)
emoji.unemojify('👍🏽')          // ':+1:'  (recognised)
emoji.strip('X👍Y')             // 'XY'    (plain emoji works)

char-regex already tokenises 👍🏽 as one character, so the gap is purely in the lookup. The README documents strip as "Remove all of the emojis from a string" and replace as "Replace the emojis in a string", and a skin-toned emoji is unambiguously an emoji (the lib's own which returns +1 for it), so leaving it in place is a contract/consistency violation.

Cause

replace() looks the character up with findByCode(character) on the raw character, while which() first normalises the skin tone:

// which.ts
const result = findByCode(skinTone(emoji, 'none'))
// replace.ts (before)
const found = findByCode(character)

strip() delegates to replace(), so both are affected.

Fix

Apply the same skin-tone normalisation in replace() that which() already uses. skin-tone is already a dependency, and the base shortcode table contains no skin-tone scalars, so plain emoji and existing matches are unaffected.

Tests

Added skin-tone cases to strip.test.ts and replace.test.ts. They fail on main and pass with the fix. Full suite green (74 tests), including the existing complex-ZWJ, flag, and preserveSpaces cases.