CSRMatrix::laplacianMatrix wrong results for self-loops

[bernlu]

The laplacian matrix computed by CSRMatrix::laplacianMatrix is wrong when the graph contains self-loops.

Example:

    Graph G(2);
    G.addEdge(0, 1);
    G.addEdge(1, 1);

    std::cout << CSRMatrix::laplacianMatrix(G) << std::endl;
    // 1, -1,
    // -1, -1,

I am not 100% sure what the expected result is, but in any case it is not -1 on the diagonal.

[hmeyerhenke]

My suggestion: Test if there are self-loops. Exclude/ignore them. Document this in the code and tell the calling function that self-loops were ignored.