Fix eigen copying of non-standard stride values

Some Eigen objects, such as those returned by matrix.diagonal() and
matrix.block() have non-standard stride values because they are
basically just maps onto the underlying matrix without copying it (for
example, the primary diagonal of a 3x3 matrix is a vector-like object
with .src equal to the full matrix data, but with stride 4).  Returning
such an object from a pybind11 method breaks, however, because pybind11
assumes vectors have stride 1, and that matrices have strides equal to
the number of rows/columns or 1 (depending on whether the matrix is
stored column-major or row-major).

This commit fixes the issue by making pybind11 use Eigen's stride
methods when copying the data.

example/eigen.cpp

@@ -56,6 +56,16 @@ void init_eigen(py::module &m) {
     m.def("cholesky5", &cholesky5);
     m.def("cholesky6", &cholesky6);
 
+    // Returns diagonals: a vector-like object with an inner stride != 1
+    m.def("diagonal", [](const Eigen::Ref<const Eigen::MatrixXd> &x) { return x.diagonal(); });
+    m.def("diagonal_1", [](const Eigen::Ref<const Eigen::MatrixXd> &x) { return x.diagonal<1>(); });
+    m.def("diagonal_n", [](const Eigen::Ref<const Eigen::MatrixXd> &x, int index) { return x.diagonal(index); });
+
+    // Return a block of a matrix (gives non-standard strides)
+    m.def("block", [](const Eigen::Ref<const Eigen::MatrixXd> &x, int start_row, int start_col, int block_rows, int block_cols) {
+        return x.block(start_row, start_col, block_rows, block_cols);
+    });
+
     m.def("fixed_r", [mat]() -> FixedMatrixR { 
         return FixedMatrixR(mat);
     });