bug in radon transform?

[Emmanuelle Gouillart]

	Hello,

	I have the impression that the center of rotation used for
computing the radon transform (skimage.transform.radon) is not correct:

>>> import numpy as np
>>> from skimage.transform import radon, iradon
>>> 
>>> a = np.zeros((3, 3))
>>> a[1, 1] = 1
>>> projs = radon(a, [0, 45, 90, 135])
>>> projs
array([[ 0.        ,  0.        ,  0.        ,  0.        ],
       [ 0.        ,  0.        ,  0.        ,  0.13603897],
       [ 0.        ,  0.46446609,  1.        ,  0.79289322],
       [ 1.        ,  0.46446609,  0.        ,  0.        ],
       [ 0.        ,  0.        ,  0.        ,  0.        ],
       [ 0.        ,  0.        ,  0.        ,  0.        ],
       [ 0.        ,  0.        ,  0.        ,  0.        ]])

(the projection at 90�C, that is projs[2], should have the non-zero
coefficient centered). 

Before I try to modify this, I'd like to be sure that it is not the
expected behaviour, and a bug indeed... 

	Cheers,

	Emmanuelle