Graphscraper

Fixed a bug in NodeList and improved the handling of the Spotify API.

Changes:
- Fixed the neighbor count limiting feature of `SpotifyArtistNode`.
- Made the neighbor count limitation configurable in `SpotifyArtistGraph`

No real changes. The main reason for creating this release is to prepare for the registration of the project on pypi.

First release of the project.

Contents
- Base graph (`graphscraper.base.Graph`) implementation with _SQLAlchemy_ and _Flask-SQLAlchemy_ database interface (`graphscraper.db`) and automatic database management.
- An example `Graph` implementation (`graphscraper.igraphwrapper.IGraphWrapper`) that uses an `igraph.Graph` instance (see the _igraph_ project) as the "remote" data source.
- A demo (`graphscraper.demo`) showing the basic usage of the _graphscraper_ project.

* Fixed some caching issues.
* Added an `external_id` property to `Node`.
* Added a Spotify API based graph scraper implementation where artists are the nodes of the graph and two artists are connected if they are related according to the Spotify API related-artists endpoint.

For the individual requirements of specialised graph implementations please take a look at the corresponding module's documentation.