If a star gravity can change the apparent location of another star on the sky, can it change the gravitational field vector of that star? If a star gravity can change the  apparent location of another star on the sky can it change the gravitational field vector of the second star? In simple words if a star gravity deflects the position of second star would an observer measure gravitational pull of the second star from its real position or from the position caused by the light deflection?
 A: The answer to the question in the title is yes, because general relativity is background independent. The gravitational influence of a body is directed toward the body and falls of with distance, and the only reference for direction and distance is the gravitational field (spacetime) itself, so it's impossible for the gravitational field of one body to ignore the field of another.
I wouldn't expect the effect to be the same as the bending of starlight, because it doesn't usually make sense to treat the field (as opposed to "updates" to the field, i.e. gravitational waves) as propagating at the speed of light. Even without an intervening body, the tidal acceleration of the Sun doesn't point in quite the same direction as the light from the Sun, because the latter is subject to aberration and the former isn't.
I'm not sure how to calculate the size of the effect. The calculation of the bending of starlight treats the light as a test particle with negligible mass, but you can't do that here because both of the bodies have to have a nontrivial gravitational field. You also can't use linearized GR because any interaction between the fields of two bodies is nonlinear by definition. Modeling the lensing body with the Schwarzschild geometry and the other gravitating body as a perturbation on top of that might work.
A: 
If a star gravity can change the apparent location of another star on the sky, can it change the gravitational field vector of that star?

You are confusing the collective effect of the gravity of a given star on electromagnetic waves,  to the gravity of the stars whose images the electromagnetic waves carry when  we detect the lensing. The star (or  galaxy) whose image is  carried by the electromagnetic wave passing another star and undergoes lensing, is many light years away and the gravitational attraction between the focus star and the one giving the image is zero at such distances.
It is called lensing because it works as distorted images through in-homogeneous glass.


if a star gravity deflects the position of second star would an observer measure gravitational pull of the second star from its real position or from the position caused by the light deflection?

In the case of lensing by the sun, which has been measured, it was measured by waiting for a total eclipse, so that the brightness of the sun would not obliterate the images. As the constellations and stars can be measured when there is  no sun in front, it is easy to see what the image distortion due to the gravitational field of the sun  is, and it checks with the calculations.
One might be able to estimate the gravitational field of the stars in the lensing image  if they are part of galactic trajectories, and this should be the same whether estimated in the lensed  image or directly, although the mathematics for the distortion of the images would be very complicated to do.
This article may help.
