[FIX] Add emg to timeseries rule for physio/stim files

[neuromechanist]

Summary

• Add emg to the base timeseries rule datatypes in src/schema/rules/files/raw/task.yaml
• This allows _physio, _physioevents, and _stim suffixes to be valid within emg/ directories
• EMG was the only datatype missing from the timeseries rule, despite being a first-class BIDS datatype since v1.11.0 (BEP 042)

Context

When converting the HySER dataset to BIDS-EMG format, finger force sensor recordings (_physio.tsv.gz) placed alongside EMG data cause ALL_FILENAME_RULES_HAVE_ISSUES validation errors because the validator schema does not include emg in the timeseries rule.

All other electrophysiology datatypes (eeg, ieeg, nirs) as well as beh, func, meg, motion, pet, anat, dwi, and perf already support physio/stim files.

Test plan

• Verify emg is a defined datatype in src/schema/objects/datatypes.yaml
• Run schema tests
• Validate a BIDS-EMG dataset with _physio.tsv.gz files passes after this change

[effigies]

I agree with this. A question is whether this should wait on (or justify) another BIDS release (1.11.1) or can be a schema patch (schema-1.2.1) on v1.11.0.

For it to be a schema patch, it cannot change the semantic meaning of the BIDS specification version 1.11.0. While I think this was obviously intended in v1.11.0, it does change the literal text of the document, so I think a stickler could argue for a new spec release.

There has been very little activity since the 1.11.0 release, so we could simply make a 1.11.1 release with this fix, and avoid the whole question.

Opinions? @bids-standard/maintainers @bids-standard/steering?

---

Edit: Per discussion with BIDS maintainers, we will release a 1.11.1 without the usual 1 week notice, as the change is so small.

[effigies]

@neuromechanist Could you add a physio file to an example dataset to validate this?

[bendhouseart]

👍 lgtm