Our current implementation is manual and verbose and can be replaced with `np.array_equal` along with the `equal_nan=True` argument