The standard for NTSC is that the 8 outermost pixels should be blank and are meant for cropping and not be displayed. The centre 704 pixels contains the video and determine the AR. These are displayed as 640x480 square pixels for a 4:3 AR. So if you remove the black bars and rescale the GOUT to it's correct AR, you get a 640x274 video with an AR of 2.34:1. If you then upscale by a factor of 2, you have a video of 1280x548. Now as you have noticed the blank areas of the GOUT are not entirely blank. They contain video. However if you were to upscale the 714x274 video to 1280x548 (the left 4 pixels and the right 2 pixels are blank for Star Wars) you would get the wrong AR, since the true AR is determined by the centre 704 pixels, not 714 pixels. So I did what was necessary to keep the correct AR and that is the crop the video to 704x274 and upscale to 1280x548.

It was CatBus who provided this information thankfully. So all credit should go to him. I didn't know about this, but was struggling with getting the correct AR.