* arxiv2bib/arxiv2bib.py: Remove the for loop over all arxiv
  numbers. Instead, merge them in a single query to inspire. This as
  two advantages. First, there is minor danger to overload inspire
  servers. Second, bibtex retrieval is much faster.

* README: Update according to the change above.

* setup.py: Require `future` module to be installed. It has been
  introduced for compatibility with both Python 2 and 3.