Arvados

It would be very helpful to have hashes for the individual files in a collection available by default. This supports a couple of use cases:

1. Checking if a file in Keep is the same as a local file (even if they have different file names) for validation, or to avoid redundant file uploads.
2. Searching to see if a given file is present in Keep (for example, auditing to see if a file was uploaded that should not have been.)
3. Identifying files which are stored multiple times in Keep with different block alignment (resulting in different blocks) which could be re-packed and the relevant collections updated to achieve deduplication).

Implementation proposal sketch here:

https://arvados.org/projects/arvados/wiki/Separating_files_from_collections