Bug in bulk operations

[andryw]

The bulk helper from python elasticserach library receives "raise_on_error" and "chunk_size" parameters.
The bulk operations will be executed for "chunk_size" documents at time (default 500).
If raise_on_error is True, an exception is raised and the documents not processed at this point will not be processed.
This is problematic in delete_all operation of ESEngine.

If I send 501 documents to be deleted, which the first 500 documents aren't idenxed but the 501° is, an exception will be raised on the first "chunck" and the 501° will not be deleted.

I know that it's possible to pass kwargs on esengine. Perhaps this behavior could be more explicit.
What do you think?