I agree with your location contender: Google street view has recent pass in May 2022. I believe the two small, triangular lighter colored trees next to the two taller darker ones with exposed trunkes visible in the article photo are visible from here: https://goo.gl/maps/yXNWpyqy4RgpUChC7
Because the "two triangluar trees" in the photo are actually three trees (observe them from further north on southbound lane), we can line up the direction the photo was taken from almost exactly. The photo must almost certainly be on a line from the triangluar trees through here: https://goo.gl/maps/snxfWCeMXeKoNdWJ9
(Google street view from the nearby of the southbound rest stop/lot itself -ie not from the southbound lane- is dated from 2011 so the trees are different)
My own speculation:
If that is the case, and given the history of "autopilot" being confused about white lines indicating that the car must surely be in the lane (eg driving right into wedges at offramps), It's possible the 'double offramp' design of the southbound parking lot/rest stop here would be particularly challenging for an automated system.
Because the "two triangluar trees" in the photo are actually three trees (observe them from further north on southbound lane), we can line up the direction the photo was taken from almost exactly. The photo must almost certainly be on a line from the triangluar trees through here: https://goo.gl/maps/snxfWCeMXeKoNdWJ9
(Google street view from the nearby of the southbound rest stop/lot itself -ie not from the southbound lane- is dated from 2011 so the trees are different)
My own speculation: If that is the case, and given the history of "autopilot" being confused about white lines indicating that the car must surely be in the lane (eg driving right into wedges at offramps), It's possible the 'double offramp' design of the southbound parking lot/rest stop here would be particularly challenging for an automated system.