fix(eyetracking): read eyelink calibration files as UTF-8 (#14000)

[CedricConday]

Reference

Closes #14000

What was wrong

mne.preprocessing.eyetracking.read_eyelink_calibration() decoded the .asc file as ASCII:

lines = fname.read_text(encoding="ASCII").splitlines()

When a MSG line contained a non-ASCII character (e.g. a zero-width no-break space, U+FEFF, which encodes to the 0xef byte from the traceback), this raised UnicodeDecodeError and the whole read failed. As the reporter notes, MSG lines are not guaranteed to be ASCII — users may put UTF-8 text in their messages — and mne.io.read_raw_eyelink() already reads the same files without issue.

Fix

Decode as UTF-8 instead. ASCII is a subset of UTF-8, so existing all-ASCII files are unaffected; this only widens what can be read. This matches the maintainer's suggestion in the issue thread ("I think we should decode as UTF8. If later we need to add a param to control this we can").

Test

Adds test_calibration_non_ascii, mirroring the existing test_calibration_newlines: it injects a MSG line containing non-ASCII content (including U+FEFF, the exact character that triggered the original crash) into the testing .asc, then asserts the calibration reads without error and the parsed values are unchanged.

[larsoner]

Pre-commit CI failed

codespell............................................................................Failed
- hook id: codespell
- exit code: 65

mne/preprocessing/eyetracking/tests/test_calibration.py:302: caf ==> calf

can you either modify the line or add caf to ignore_words.txt?

[CedricConday]

Thanks! Fixed in 2565238 — reworded the test message to drop the caf token (kept the U+FEFF character that triggers the original crash). pre-commit is green now.

[larsoner]

Marking for merge when green, thanks in advance @CedricConday !

[welcome]

🎉 Congrats on merging your first pull request! 🥳 Looking forward to seeing more from you in the future! 💪