[FIX] restructure and clarify *_physio/*_stim section

[sappelhoff]

Today I needed to save some *_physio files for the first time and went through the BIDS specification on that topic in detail.

I think the section is currently quite disorganized and not formatted consistently with other BIDS sections.

I took a pass and performed:

• a major restructuring of the paragraphs to accommodate a better "reading flow" from a user perspective
• wording clarifications

The diff is quite large but I did not introduce new notions or change old ones (at least not to the best of my knowledge).

Please let me know if you agree that this PR improves the section ... or/and help me to make it (even) better.

[sappelhoff]

note: I should perhaps change the examples / templates from MRI centric to BIDS:

sub-<label>/[ses-<label>/]
-    func/
+    [func/ or meg/ or eeg/ or ieeg/ or beh/]
        <matches>[_recording-<label>]_physio.tsv.gz

does somebody have a prettier idea to get this point across?

[effigies]

note: I should perhaps change the examples / templates from MRI centric to BIDS:

sub-<label>/[ses-<label>/]
-    func/
+    [func/ or meg/ or eeg/ or ieeg/ or beh/]
        <matches>[_recording-<label>]_physio.tsv.gz

does somebody have a prettier idea to get this point across?

In derivatives, we've used <datatype>/ to as a general stand-in data type. If you want to be specific, you could go for <func|meg|eeg|ieeg|beh>/. I can't immediately find an example where we've done that, but I think it would be clear.

[effigies]

Is it necessary to have a header here? If so, let's think of a more descriptive name.

[sappelhoff]

I think it helps. Everything prior to this header contains the "essential rules" ... everything below contains some half-complete recommendations on specific use cases.

Hence, how about Recommendations for specific use cases?

[sappelhoff]

I think it helps. Everything prior to this header contains the "essential rules" ... everything below contains some half-complete recommendations on specific use cases.

Hence, how about Recommendations for specific use cases?

[sappelhoff]

In derivatives, we've used / to as a general stand-in data type. If you want to be specific, you could go for <func|meg|eeg|ieeg|beh>/. I can't immediately find an example where we've done that, but I think it would be clear.

I like func|meg|eeg|ieeg|beh, but given that we may want to allow it for anat and dwi as well (x-ref: https://github.com/bids-standard/bids-validator/issues/990#issuecomment-649627395), datatype may be the shorter variant, and it's good to do the same as in derivatives.

[robertoostenveld]

I consider this a good clarification of the specification.

There are a few more features/characteristics that would be good to consider (e.g. synchronization with the other data, whether it should be present in the scans.tsv file) but those fall out of scope for this improvement.