Large datalad dataset are faster and better behaving when modularized, but from my understanding, openneuro datasets are only a single datalad repo. Are there any plans to allow uploading a modular dataset under a single entry on openneuro? If a project produces many sub-datasets which would each contain derivatives sub-datasets, what would be the best way to upload it?
I know that openneuro focuses on raw-data, but standard pipeline outputs (eg. fMRIPrep) are often shipped with datasets, in a BIDS (non-YODA) way. Are there plans to migrate to a YODA approach to track provenance?