This approach would support a user uploading files as part of their dataset that contain the metadata. For each metadata format we support we would provide a dedicated API endpoint that uploads the file as unusual then validates the file according to the associated schema. When the record is published we'd open up the file and index the metadata in Opensearch.

The question with this approach is how well it would work with "remote" data files (see #281). We'd also ideally figure out how to display the contents of the metadata from the file without having to open it everytime (maybe by using the data in the index).