fix: read and write text config files as UTF-8

[yosofbadr]

Summary

Draft / waiting on broader validation.

Fixes #1345 by reading and writing TagStudio text configuration/resource files with explicit encodings instead of relying on the system default encoding.

This keeps the change focused on text encoding handling: settings are read as UTF-8 first with a fallback for legacy non-UTF-8 files, then saved as UTF-8. Related generated ignore/resource text reads and writes now also specify UTF-8. I removed an unrelated generated uv.lock change from this PR.

Tasks Completed

• Platforms Tested:
  • Windows x86
  • Windows ARM
  • macOS x86
  • macOS ARM - targeted local test/lint
  • Linux x86
  • Linux ARM
    • Not manually tested on a non-UTF-8 Windows desktop; covered locally by a cp932 regression fixture.
    • GitHub Actions for the latest push are awaiting maintainer approval on the fork.
• Tested For:
  • Basic functionality - targeted settings encoding regression tests
  • PyInstaller executable
    • Local: uv run --extra pytest python -m pytest tests/qt/test_global_settings.py -q
    • Local: uv run --extra ruff ruff check src/tagstudio/qt/global_settings.py src/tagstudio/core/library/refresh.py src/tagstudio/qt/resource_manager.py src/tagstudio/core/library/alchemy/library.py tests/qt/test_global_settings.py
    • Missing before marking ready again: latest GitHub Actions run and/or Windows non-UTF-8 codepage validation.

[yosofbadr]

Closing this after a stricter audit. The issue is Windows/codepage-specific and the PR only has a synthetic local fixture, not real Windows validation. I do not want to leave maintainers with a PR I cannot confidently stand behind.

Sorry for the noise.