expose itemsize and nbytes as for numpy arrays

see:
  * https://numpy.org/doc/stable/reference/generated/numpy.ndarray.nbytes.html
  * https://numpy.org/doc/stable/reference/generated/numpy.ndarray.itemsize.html

relates to ml-explore/mlx-examples#174

python/src/array.cpp

@@ -510,6 +510,14 @@ void init_array(py::module_& m) {
           "size", &array::size, R"pbdoc(Number of elments in the array.)pbdoc")
       .def_property_readonly(
           "ndim", &array::ndim, R"pbdoc(The array's dimension.)pbdoc")
+      .def_property_readonly(
+          "itemsize",
+          &array::itemsize,
+          R"pbdoc(The size of the array's datatype in bytes.)pbdoc")
+      .def_property_readonly(
+          "nbytes",
+          &array::nbytes,
+          R"pbdoc(The number of bytes in the array.)pbdoc")
       // TODO, this makes a deep copy of the shape
       // implement alternatives to use reference
       // https://pybind11.readthedocs.io/en/stable/advanced/cast/stl.html

python/tests/test_array.py

@@ -84,6 +84,8 @@ def test_array_basics(self):
         x = mx.array(1)
         self.assertEqual(x.size, 1)
         self.assertEqual(x.ndim, 0)
+        self.assertEqual(x.itemsize, 4)
+        self.assertEqual(x.nbytes, 4)
         self.assertEqual(x.shape, [])
         self.assertEqual(x.dtype, mx.int32)
         self.assertEqual(x.item(), 1)