Could consider putting it up as a dataset on Kaggle, perhaps? I would think they'd provide hosting for such things?

Archive.org would be another option as a repository for the high-res scans in an accessible / discoverable location.