min and max gives arbitrary result (instead of nan) when there is nan #87

In [1]: import riptable as rt

In [2]: rt.version
'1.0.38'

In [3]: import numpy as np

In [4]: ar = np.array([1.0, np.nan, 3.0])

In [5]: np.min(ar)
nan

In [6]: tmp = rt.Dataset({'f': ar})

In [7]: np.min(tmp['f'])
3.0

In [8]: tmp['f'].min()
3.0