Improve large file upload resiliency

Improvement 1

EDIT: we had issues impacting the FM call, but it was not a timeout one. So keeping as-is as it works fine for now.

When completing an upload, the FileManager call is done with a big timeout (21,600 seconds). It is so to handle big files.
Let's see what can be done to get rid of that huge timeout.

Improvement 2

If an error happens at the call to the FileManager, Drive will retry 60 seconds later. But then, as the batchId was used once, it has been automatically deleted: so a new upload will start from ground.

The remote client should make use of the X-Batch-No-Drop header to prevent the batchId purgation after its first use. And a manually cleanup should be done only when the upload is fully done.