Keep existing repr() results, with numpy>=2.0

In various __repr__() implementations, replace 'list(ndarray)' with
'ndarray.tolist()', in order to avoid inclusion of type information (like
'np.int64(1)') by the latest numpy version.

This also fixes test_clifford.TestPrettyRepr and test_transformations
pytest-s, as they rely on hardcoded results from IPython.lib.pretty.

Note: numpy>=2.0 is allowed by numba>=0.60.

clifford/_layout.py

@@ -479,7 +479,7 @@ def vee(aval, bval):
     def __repr__(self):
         return "{}({!r}, ids={!r}, order={!r}, names={!r})".format(
             type(self).__name__,
-            list(self.sig), self._basis_vector_ids, self._basis_blade_order, self.names
+            self.sig.tolist(), self._basis_vector_ids, self._basis_blade_order, self.names
         )
 
     def _repr_pretty_(self, p, cycle):
@@ -489,7 +489,7 @@ def _repr_pretty_(self, p, cycle):
         prefix = '{}('.format(type(self).__name__)
 
         with p.group(len(prefix), prefix, ')'):
-            p.text('{},'.format(list(self.sig)))
+            p.text('{},'.format(self.sig.tolist()))
             p.breakable()
             p.text('ids=')
             p.pretty(self._basis_vector_ids)

clifford/_multivector.py

@@ -579,7 +579,7 @@ def _repr_pretty_(self, p, cycle):
                 p.pretty(self.layout)
                 p.text(",")
                 p.breakable()
-            p.text(repr(list(self.value)))
+            p.text(repr(self.value.tolist()))
             if self.value.dtype != np.float64:
                 p.text(",")
                 p.breakable()