Add new features from igraph

[thomasp85]

These features goes up to igraph 1.5.1

• largest_component()
• ivs(), largest_ivs(), maximal_ivs(), ivs_size()
• weighted cliques
• Pertaining to above, find a way to wrap algorithms where nodes can be part of several groups (e.g ivs(), cliques())
• reverse_edges()
• eulerian_path(), eulerian_cycle() + has_ variants
• sample_spanning_tree()
• random_walk()/random_edge_walk()
• harmonic_centrality()
• greedy_vertex_coloring()
• efficiency algorithms
• feedback_arc_set()
• cluster_leiden() and cluster_fluid_communities()

[szhorvat]

• maximal_ivs() - will soon be renamed to max_ivs() (in 1.6.0)
• weighted cliques: just be aware that only integer weights are supported at the moment, and this is likely to stay this way for a long time
• "find a way to wrap algorithms where nodes can be part of several groups" 👍
• random_walk()/random_edge_walk() In the C library these two were unified. Not sure what will happen in R. But one result of the unification was that the number of steps is specified differently. Right now, in 1.5.x, steps specifies the number of vertices to return, not the number of steps to take. From 1.6.0, this will truly be the number of steps. So random_walk(g, v, n) will take n steps, and will therefore return n+1 vertices, including the starting point. random_edge_walk() already does the right thing.
• cluster_leiden() — be aware of cluster_leiden() docs should be explicit about the objective function igraph/rigraph#529, not sure what the outcome will be. Also be aware of cluster_leiden() resolution_parameter should be renamed to resolution igraph/rigraph#883

I hope this helps making decisions about which to include now and which to postpone for later.

[thomasp85]

Appreciate the input @szhorvat