add support for pd.Series in asarray

Today I was bitten/surprised by the fact that la.asarray(Series) works but drops the index.

```python
>>> asarray(pd.Series([0, 1, 2], ['a0', 'a1', 'a2']))
{0}*  0  1  2
      0  1  2
```
In fact, it's just that Series are not handled in asarray, so this goes via the `__array__` path.