[Bug] Surrogate pair handled incorrectly

[exoego]

Describe the bug

Surrogate pairs render brokenly, because fontSubstitution walks the string one UTF-16 code unit at a time and queries the high and low surrogate halves against hasGlyphForCodePoint separately, so the combined code point is never looked up in the font stack.

Affected characters

• Font-based Emoji
• Special Kanji. Unicode SIP (Supplementary Ideographic Plane) characters (e.g. 𠮷 U+20BB7),
• Egyptian Hieroglyphs
• Cuneiform
• Mathematical Alphanumeric Symbols
• etc

To Reproduce
playgorund

Expected behavior
fontSubstitution should combine each surrogate pair into its single code point before consulting the font stack, so a font that supports surrogate pair (e.g. Noto Sans JP for U+20BB7) is picked and the glyph renders correctly.

Desktop (please complete the following information):

• OS: Linux
• Browser: Node.js
• React-pdf version: v4.5.1