Implement a parallel way to zip Blobs with bulk service

We would like to parallelize the blobs retrieval when Nuxeo Server is asked to zip them.

This is currently done sequentially by the BlobUtils#zip method (for instance in BulkDownload operation) or by the DocumentReader/DocumentWriter API (for instance in ExportDocument operation).

We could replace/improve the BulkDownload operation with a BAF action (such as the one done for CSVExport) with a computation doing the download in parallel.

For other cases, such as DocumentReader/DocumentWriter, as the operation could be included in other chains, we may want to provide a solution with a dedicated Executor.