Create new BookstoreLargeContentsArchiver for large contents

[mpacer]

In the classic notebook server, we have the LargeFileManager to handle large files as a streamed response. We could have something analogous for bookstore that would alleviate issues around archiving large files.

There are a few details that would need some explicit design work (e.g., do we also stream responses back to S3, do we queue a save in the case where saving to S3 is taking a long time instead of debouncing all requests, &c. ), but this seems well within scope of this project.