When creating a GridFS download stream, we set the expected chunk size to be the bucket's chunk size. This isn't correct though because the chunk size can be overridden for individual files, so we should instead set it to the chunkSize field in the files collection document.