Declare nemar_id for 15 datasets whose NEMAR deposits are now public

[bruAristimunha]

Follow-up to #1139. That PR added the NEMAR download provider; this makes it actually reach 15 more datasets.

Each of these already carried a placeholder:

# nemar_id = "nm000174" pending: NEMAR deposit not yet public

Those deposits are public now and publish sourcedata/, so the placeholders are simply uncommented — 15 lines, no logic change.

Why it matters

Without the attribute the provider cannot see them. sourcedata_path() raises declares no nemar_id, and "auto" silently falls back to the upstream host — which for several of these is precisely the slow or bot-gated source the mirror exists to route around.

Measured before this change: 120 enriched deposits were reachable, 15 were not. All 15 were enriched in the last few days, after the mapping was last updated.

Verification

Every pair checked against the live deposit, not a local table. For each: the deposit is published, carries sourcedata/, and its sourcedata_provenance.json records the same MOABB class. This mattered — two entries in our own mapping table were wrong in exactly this way (Chailloux2020 pointed at a duplicate of on003190, TrianaGuzman2024 at a deposit the NEMAR admins had denied), so checking the table against itself would have propagated those errors.

End-to-end with the provider pinned to nemar:

dataset	deposit	files verified	mismatched
Nguyen2017_V	nm000261	8	0
Pressel2016	nm000258	15	0
Kaneshiro2015	nm000263	10	0

Nguyen2017_V previously failed with declares no nemar_id; it now downloads. Every file hashes to the sha256 recorded from the original upstream fetch at staging time, so the served bytes equal the upstream bytes.

Separately, a checksum cross-check over a 20-deposit sample compared 2,719 files between each published manifest and its provenance — zero mismatches.

Note for a follow-up (not this PR)

nemar-py logs primary backend failed; falling back to next layer on every fetch, because small git-keyed files (README.md, sourcedata_provenance.json) have no annex checksum for the S3 backend to use. It recovers correctly, but a routine fallback reading as a failure makes real failures harder to spot.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.