scipy: `interp1d` behaviour inconsistent with Numpy's?

Scipy interpolator returns nan if one of the neighbor is a nan even when interpolation is not actually needed. This looks inconsistent with Numpy’s behaviour :

>>> import numpy as np
>>> from scipy.interpolate import interp1d
>>> np.interp([0,1,2], xp=[0,1,2], fp=[np.nan,0,1])
array([nan,  0.,  1.])
>>> interp1d([0, 1, 2], [np.nan, 0, 1])([0,1,2])
array([nan, nan,  1.])

Is that an expected behaviour ?

Thanks a lot,

Version information :

>>> import sys, scipy, numpy; print(scipy.__version__, numpy.__version__, sys.version_info)
1.6.0 1.19.4 sys.version_info(major=3, minor=8, micro=5, releaselevel='final', serial=0)