Add Unicode-aware line wrapping with conservative number protection

[nguillot]

Summary

Implements intelligent multi-line text wrapping with proper width constraint enforcement and international language support. The implementation uses Core Text's Unicode Line Breaking Algorithm with a conservative validation layer to protect numbers from being split across lines.

Problem

Previously, when setting preferredMaxLayoutWidth on MTMathUILabel, long text would either:

1. Overflow beyond the width constraint - Text like "Rappelons la conversion : 1 km équivaut à 1000 m." would render at 262pt even with a 235pt
   constraint, causing truncation
2. Break incorrectly - Simple space-based breaking didn't handle:
   - CJK languages (Chinese, Japanese) where words aren't space-separated
   - Number formats across locales (French 3,14, English 1,000,000, French 1 000 000)
   - Complex grapheme clusters (emoji, combining characters)

Example of the bug:

Constraint: 235pt
Actual width: 262.786pt
Rendered: "Rappelons la conversion : 1 km équi[TRUNCATED]"
expected: "Rappelons la conversion : 1 km équivaut à 1000 m."

Implementation Details

Number Protection Patterns

The implementation conservatively protects these number formats:

Format	Example	Protected?	Use Case
French decimal	3,14	✅	European decimals
English decimal	3.14	✅	US/UK decimals
Thousands (comma)	1,000,000	✅	English numbers
French thousands	1 000 000	✅	French numbers
Swiss thousands	1'000'000	✅	Swiss numbers
Scientific notation	6.022e23	✅	Scientific text

Wrapping Behavior

CJK Languages:

Chinese: Can break between most characters
Japanese: Follows kinsoku rules (禁則処理)
Korean: Breaks at spaces (like English)

Latin Languages:

English/French/Spanish/German: Break at word boundaries (spaces)
Respects accented characters: é, ñ, ü, etc.

Mixed Content:

"The equation is 方程式: x² + y² = r²"
→ Can break between scripts intelligently