Most minor observation ever: I think perhaps you don't quite have the correct output aspect ratio. Pixels in Mode 1 aren't quite square. Pixels run for 40/64ths of the line, which is close to 77% of the visible area. 320 pixels of height would be about 111% of the visible area. So I make each Mode 1 pixel only about 92% as wide as tall.
EDIT: proper workings:
Per PAL, 52us of each line is visible and 288 lines are visible, with the screen having an aspect ratio of 4/3. In Mode 1 there are 320 pixels across and they take 40us to output, with each line being unique.
So each pixel is 1/288th of the screen tall and (40/52) * (1/320) = 1/416th of the screen wide. Correcting for screen aspect ratio, each pixel has an aspect ratio of (288*4)/(416*3) = 12/13.
So, a touch thinner than tall.