Use multipart API for downloads
Currently we use the upload_fileobj
method for uploads which handles
switching to the multi-part API for large blobs (the comment was wrong).
For downloads we previously used a simple get.
Update the download logic to use the multi-part download API download_fileobj
to take advantage of threads and parallelism when blobs are large. Currently
use the default configuration (in the future we might want to make that extensible).
Edited by Ed Baunton