Call it at the point the bytes come down (urls: method download,
accessed now) or, for archives kept on Drive, on the files themselves
(files: method file_mtime), and hand the result to
build_metadata_json()'s sources argument so it lands in the ingest's
metadata.json as sources[]. The release then takes the newest stamp as
the dataset's source_accessed (resolve_source_accessed()).
Value
A tibble: source, method, accessed (POSIXct,
UTC), bytes.
sources_block(): the list build_metadata_json() writes as
sources[].