comparison between Array and None (or an Axis) returns False (single value)

This is a regression in 0.33 (39ced81f9a335f47bbf512c50139e35f1ff86705) because None is not a scalar for np.isscalar

Test case:
```python
>>> arr = full("a=a0..a2", None)
>>> arr == None
False
```
One workaround is (we can use `is` too but it is farther from the original code):
```python
>>> arr.apply(lambda v: v == None)
a    a0    a1    a2
   True  True  True
```
