`points` tuple) must be
    strictly ascending or descending.      
values : ndarray of float, shape (m1, ..., mN, ...)
    The data on the regular grid in n dimensions.
k : int, optional
    The spline degree. Must be odd. Default is cubic, k=3
solver : a `scipy.sparse.linalg` solver (iterative or direct), optional.
    An iterative solver from `scipy.sparse.linalg` or a direct one,
    `sparse.sparse.linalg.spsolve`.
    Used to solve the sparse linear system
    ``design_matrix @ coefficients = rhs`` for the coefficients.
    Default is `scipy.sparse.linalg.gcrotmk`
solver_args : dict, optional
    Additional arguments for the solver. The call signature is
    ``solver(csr_array, rhs_vector, **solver_args)``

Returns
-------
spl : NdBSpline object

Notes
-----
Boundary conditions are not-a-knot in all dimensions.
c