test(nemar): drop 15 published deposits from NEMAR_ID_PENDING

[bruAristimunha]

Summary

Drop 15 NEMAR deposits from NEMAR_ID_PENDING in moabb/tests/test_datasets.py so that test_all_datasets_have_valid_nemar_id asserts their published ids instead of skipping them. These 15 deposits were pushed to the admin-published state and became anonymously reachable on NEMAR; the pending list on develop was lagging behind reality.

No loader scientific change. Every one of these 15 loaders already declared its nemar_id on develop, so the NEMAR download-provider path was already active for them. The only visible change is that the regex validity test now runs against these class attributes instead of being skipped. No changelog entry (test-only; please apply the no changelog label).

Evidence

Verified from Voyager's login node on 2026-10-05, one id at a time:

1. nemar dataset status <id> → Status: active, confirming the deposit record.
2. GIT_CONFIG_GLOBAL=/dev/null GIT_TERMINAL_PROMPT=0 git -c credential.helper= clone --depth 1 https://github.com/nemarDatasets/<id> /tmp/w41-anon/<id> → success (public anonymous clone, no credentials).
3. For one annexed EEG file, resolved its S3 location via git annex whereis (nemar-s3 publicurl), then curl -sI https://nemar.s3.us-east-2.amazonaws.com/<id>/objects/<key> → HTTP/1.1 200 OK.

Full report: .pr-audits/w41/p2-evidence/anon-report.tsv (local; not committed).

NEMAR id	Loader class	publish	anon clone	annexed EDF (HEAD)
nm000174	AguileraRodriguez2025	active, published 2026-06-22	ok	200 (sourcedata/.../file_downloaded.edf, 13.3 MB)
nm000177	Kumar2024	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 5.5 MB)
nm000178	BNCI2020_001	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 74 MB)
nm000184	BCIComp2020WalkingERP	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-p300_run-0_eeg.edf, 3.3 MB)
nm000220	Beetl2021_A	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._acq-finaltest_run-1_eeg.edf, 50 MB)
nm000224	Nguyen2017_SL	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 44 MB)
nm000233	BCIComp2020UpperLimb	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 19.8 MB)
nm000252	Nguyen2017_L	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 44 MB)
nm000257	Nguyen2017_S	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 66 MB)
nm000258	Pressel2016	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 29.9 MB)
nm000261	Nguyen2017_V	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-imagery_run-0_eeg.edf, 66 MB)
nm000263	Kaneshiro2015	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-p300_run-0_eeg.edf, 49 MB)
nm000269	Mainsah2025_A	active, published 2026-06-23	ok	200 (sourcedata/A_01/.../A_01_SE001_CB_Test06.edf, 8.1 MB)
nm000273	Lee2019_SSVEP	active, published 2026-06-23	ok	200 (sub-1/ses-0/eeg/..._task-ssvep_acq-train_run-1_eeg.edf, 194 MB)
nm000274	Beetl2021_B	active, published 2026-06-23	ok	200 (sub-4/ses-0/eeg/..._acq-finaltest_run-1_eeg.edf, 10.3 MB)

Not in this PR

• nm000288 Ma2022 (PR Add Ma2022 (SHU) cross-session motor imagery dataset #1178): nemar dataset publish status nm000288 returned Status: requested and the anonymous https://github.com/nemarDatasets/nm000288 HEAD returned 404 on 2026-10-05 — still blocked on admin publication, left pending.
• 30 other deposits in NEMAR_ID_PENDING whose publication is still requested/none on NEMAR are unchanged.

Tests

Local, in a fresh worktree off origin/develop @ d7e0951:

PYTHONPATH=$PWD .venv/bin/python -m pytest moabb/tests/test_datasets.py \
    ::Test_Datasets::test_all_datasets_have_valid_nemar_id \
    ::Test_Datasets::test_nemar_id_class_attributes \
    --basetemp=/tmp/p2-pt -p no:cacheprovider
→ 166 passed, 40 skipped (pending publication / exempt) in 0.21s

python -m pytest moabb/tests/test_bnci.py --basetemp=/tmp/p2-pt -p no:cacheprovider
→ 8 passed

pre-commit run --files moabb/tests/test_datasets.py
→ rc=0 (all hooks Passed, non-applicable hooks Skipped)

test_download_falls_back_from_nemar and test_download_without_nemar_id_uses_fallback fail identically with and without this patch on develop — they rely on a per-process download-provider state that leaks from other tests (get_download_provider() == 'nemar') and raise NemarDownloadError before monkeypatching takes effect. Pre-existing; not touched here.