fix(ci): restore native test setup and lint gates

[lukisch]

The eight native CI cells stop before testing: six Ruff findings fail lint, the new asset test imports an undeclared Pillow dependency, and the security contract imports tomllib directly despite supported Python 3.10.

Narrow three error handlers to expected filesystem/data/Tk errors, format test imports and remove one redundant encoding header. Add the Python 3.10 tomli fallback (already supplied by pytest). Declare Pillow>=12.3.0 in dev tooling and install it in CI for image validation. Runtime dependencies, product version, lint rules, assertions and the existing eight-cell matrix remain unchanged. Document the repair under Unreleased.

Validation on Windows Python 3.12 with inherited PYTHONIOENCODING=utf-8:

• Base Ruff reproduces six findings; complete Ruff passes after repair.
• Base suite: 264 passed, 3 symlink-permission skips. Full suite after source repair: 263 passed, 4 skips; the extra skip is an existing Tk interaction test when this host cannot load its Tcl runtime. Native CI must establish that case.
• Ten fixture checks preserve malformed JSON, invalid UTF-8, denied-file and report-save error handling without a real GUI.
• After dependency declarations: all 39 metadata/security/asset tests pass. TOML/YAML and git diff --check pass; no runtime dependency or matrix change.
• Exact-head Windows/Linux/macOS CI remains required. No test or lint rule was disabled.

Author model: gpt-6.1-sol (Codex). Draft pending independent review; merges disabled. Based directly on master 8f9ff11. Canonical branch and OneDrive deployment remain unchanged; no release or version bump.

[github-actions]

Welcome! 👋 Thanks for your first pull request in this repository.

A maintainer will review it soon. Please make sure:

• Your changes are tested
• Documentation is updated if needed
• The PR description explains what and why

Thanks for contributing!