Unexpected return from mp.hermite when evaluating limit

When evaluating the limit approaching infinity of the hermite polynomial of degree zero, SciPy, GSL, and even the fp version of mpmath function agree that the output should be 1. However, `mp.hermite(0, inf)` evaluates to NaN.

```
>>> fp.hermite(0, np.inf)
1.0
>>> special.eval_hermite(0, np.inf) 
1.0
>>> mp.hermite(0, np.inf)
mpf('nan')
```