Initial commit to include directed degree measures in the network class

[wbarfuss]

New methods =nsi_indegree= and =nsi_outdegree= introduced so far. Open Questions / Points to discuss

• Shall the =typical_weight= fuctionality also be available for directed degrees?
• Shall there be a =link_attribute= keyword for all the degree measues, by that including =strength= measures?
• Does it make any sense to have the =sp_nsi_diag_k= and the =sp_nsi_diag_k_inv= methods written in the code between all the degree measures? If no, shall they be moved to another sensible place?

[wbarfuss]

Hi, two things: 1) Unfortunately it turned out that it is much more suitable to work on the issue regarding Delphines paper in only one branch. So I started by now implementing the clustering motifs here as well, although the branch is called nsi_degree. I don't know if we should or can rename something? 2) pylint discovers two new "error": C: 1, 0: Too many lines in module (5670/5600) (too-many-lines) R:122, 0: Too many public methods (113/110) (too-many-public-methods) How shall we proceed with this?

[wbarfuss]

Another point: local_clustering works with directed and undirected networks by treating the directed as undirected links. nsi_local_clustering works only with undirected networks. Shall this be harmonized?

[ntfrgl]

• Everything except the t = ...; T = ... lines in the motif clustering methods can be refactored into a helper function.
• Warning thresholds can be configured in pylintrc and are astronomical already.

[wbarfuss]

I think, that would solve than issue #49

[ntfrgl]

Please use tests.test_doctests.r() in doctests with floats, see e.g. Network.local_clustering().

[wbarfuss]

I hope I didn't forget any