Fix wrong information about dtype errors in assignments

[MilanStaffehl]

Documentation reads the following under "Limitations":

dtypes are typed correctly, so the following code will in fact cause type checkers to issue and error message:

import numpy as np
from numdantic import NDArray, Shape

# This does cause a type checking error
x: NDArray[Shape[int, int], np.float32] = np.array([[1, 2], [3, 4]], dtype=np.int32) 

This is wrong. mypy does not issue an error and passes.