Fix FiniteDifferenceStencilDerivation.rotate_weights_and_apply

Recent Numpy version do not support conversion from arbitrary iterables if they have the attribute shape. The FiniteDifferncesNotebook fails on my machine since to shape (3,3,3) and sequence of length 27 don't fit together.

I guess that recent Numpy will deprecated construction from arbitray sequences (only list and tuple, and objects with __array__ are allowed).

pystencils/fd/derivation.py

+import itertools
 import warnings
 from collections import defaultdict
-import itertools
 
 import numpy as np
 import sympy as sp
@@ -184,6 +184,9 @@ class FiniteDifferenceStencilDerivation:
                 result[max_offset - direction[1], max_offset + direction[0]] = weight
             return result
 
+        def __array__(self):
+            return np.array(self.as_array().tolist())
+
         def as_array(self):
             dim = len(self.stencil[0])
             assert (dim == 2 or dim == 3), "Only 2D or 3D matrix representations are available"
@@ -205,12 +208,12 @@ class FiniteDifferenceStencilDerivation:
 
             return result
 
-        def rotate_weights_and_apply(self, field_access: Field.Access, axis):
+        def rotate_weights_and_apply(self, field_access: Field.Access, axes):
             """derive gradient weights of other direction with already calculated weights of one direction
                via rotation and apply them to a field."""
             dim = len(self.stencil[0])
             assert (dim == 2 or dim == 3), "This function is only for 2D or 3D stencils available"
-            rotated_weights = np.rot90(np.array(self.as_array()).reshape(self.as_array().shape), 1, axis)
+            rotated_weights = np.rot90(np.array(self), 1, axes)
 
             result = []
             max_offset = max(max(abs(e) for e in direction) for direction in self.stencil)