Validate description limits using UTF-8 bytes

[mohgupta-ship-it]

This pull request was prepared and posted by Codex at Mohit’s request.

Summary

• Define the serialized description limit explicitly as 1,024 UTF-8 bytes.
• Enforce the byte limit during skill frontmatter validation while retaining the existing character bound for schema compatibility.
• Cover the exact 1,024-byte boundary and a 1,025-byte multibyte overflow.

Rationale

Character counts do not reliably enforce encoded storage or transport limits. Multibyte text can fit within a character limit while exceeding the serialized byte limit.

Validation

• 37 focused schema tests passed
• Ruff passed for all changed files
• git diff --check passed