First off, you are correct about it changing the perceived locations of objects. However, the amount light is bent (or really, the curvature of spacetime since that’s what light follows) is incredibly small on interplanetary scales, much less universal scales.
But to answer your question? Astronomers are very clever. And also computers exist. We have models that can compensate for microlensing effects, and have indeed documented the shifts in perceived locations of things before. If you’ve heard of an Einstein Ring, that’s what this is. We are able to reconstruct the image and location of the lensed object fairly well.
Additionally, there’s a very famous and interesting case of this being done to confirm some aspects about spacetime geometry. There was a supernova (iirc a type 2), and we captured it in our telescopes and from historical observations. And we saw it three times. Some astronomers calculated that we should see it again, a few years later, and calculated where exactly on the sky we should see it. They did this by calculating the path light would take from the original event to get to us. Then Hubble focused on that part of the sky during the scheduled time and wouldn’t you know it, we saw it again. Precisely in the window they calculated.
Latest Answers