So, now for something interesting... I repeated the polfilter turning angle test - but this time with the polarizing filter flipped over. The first experiments I did with the polarizing filter put on top of the spectrometer the way you would screw it onto a camera. Now I put it on "backward" - with some interesting results!
Here are the curves:
As you can see, there are differences that are not explained with noise effects. However you define the 0° angle, the curves for 0° and 180° (in the plot blue and yellow) are different from the curves for 90° and 270° (in the plot red and green).
You can see it in all channels, although again, there is a lot more noise in the blue channel.
Here are the CSV's: