fix(docx): make XML validation and console output Windows-safe

[voidborne-d]

Summary

• read XML files for XSD validation with explicit encoding="utf-8"
• replace two Unicode arrow glyphs in docx validator output with ASCII ->
• add regression tests covering UTF-8 XML reads and cp1252/Windows console output

Why

Fixes #712.

The docx office scripts currently rely on the platform default text encoding in one XML validation path, which breaks on Windows when the default codec is cp1252. They also print → in two validation/repair messages, which can raise UnicodeEncodeError on cp1252 terminals.

This keeps the XML parsing path deterministic across platforms and makes the console output safe on default Windows terminals without requiring PYTHONIOENCODING=utf-8.

Testing

• python3 -m unittest tests.test_docx_windows_encoding -v