I try to cleanly separate between information that should be part of the
Kwant source distribution (that is tied to a particular version of
Kwant), and generic information that is best presented on the website.

Now we exec _kwant_version.py into a namespace and get
the version from that.

Specify only Python3 support in trove classifiers.
https://www.python.org/dev/peps/pep-0301/#distutils-trove-classification

Python 3 no longer allows implicit relative imports:
https://docs.python.org/release/3.0.1/whatsnew/3.0.html#removed-syntax
Instead of importing the version file we just parse it. By keeping
the extension ".py" we make it so that it will be distributed along
with the rest of the package.

Running 2to3 removed a line in the graphene.py tutorial script, which
broke the diff file. We manually update the diff file to fix this.

Under Python 3, `scipy.sparse.coo_matrix` fails when passed memoryviews.
We coerce the memoryviews to numpy arrays before passing them to the
`coo_matrix`. Even though this is kind of a hack it shouldn't be very
inefficient as no data is actually copied. See
https://github.com/scipy/scipy/issues/5123 for details.