fix(saxs): harden experimental data loading on Windows

[kewh5868]

Summary

This PR incorporates the safe Windows crash fixes from #125 onto the current up-to-date main baseline.

Changes included:

• Harden experimental data loading for metadata-heavy files.
• Preserve commented column headers like # q I(q) while ignoring metadata comments such as # Background offset factor: 1.0.
• Keep detected header-row metadata even when np.loadtxt can parse the file directly.
• Bound the experimental data header dialog to the available screen size.
• Wrap long file paths and preview content to avoid oversized dialogs on Windows/OneDrive paths.
• Add focused tests for comment-header parsing and dialog geometry.

Validation

• File-scoped pre-commit hooks passed.
• Targeted pytest passed:

conda run --no-capture-output -n saxshell-py312 pytest -q \
  tests/test_saxs_ui.py::test_experimental_data_header_dialog_recovers_header_rows \
  tests/test_saxs_ui.py::test_load_experimental_data_file_detects_three_column_headers \
  tests/test_saxs_ui.py::test_experimental_data_header_dialog_allows_manual_column_selection \
  tests/test_saxs_ui.py::test_experimental_data_header_dialog_geometry_is_screen_bounded \
  tests/test_saxs_ui.py::test_experimental_data_metadata_comments_are_not_used_as_columns

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 25104474b2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Stop treating any token containing 'q' as a column label

Using "q" as a substring keyword in _tokens_look_like_column_labels causes many metadata comments (e.g., lines containing words like Acquisition) to be classified as column headers. In load_experimental_data_file, this can promote a metadata comment to column_names, after which _resolve_experimental_columns may infer column indices from metadata token positions that do not exist in the numeric matrix (for example, inferring intensity/error at index 2+ for a 2-column file), raising an out-of-range ValueError for otherwise valid data files.

Useful? React with 👍 / 👎.