Processing time (as in “process”, not language) is a whole another story, yes. I’ve been struggling with that as well, as I was trying to generate ~8-12k tall images for printing in 300 DPI in a normal poster size. I think the answer is to calculate different parts of image/animation separately and then stitch it together, but I’ve never dived into that.
Also, you just gave me an idea that it would be interesting to try and “stream” image through analog synth and then reconstruct it back 
As for the books, there is an oldie but goodie: http://www.generative-gestaltung.de/ — but it’s a little bit more on a beginner’s side.
Regarding gifs: yeah, I believe some of the noise is introduced while converting the animation itself, but I remember that I was adding some dithering noise before conversion. Just for the looks, plus without noise some parts of the animation with gradients etc. look weird because of the quantization of colors.
Here are few more:




Love your gifs too, by the way. Several years ago I had a folder on my PC where I was saving all of the animations from internet with similar vibes. I love the creepiness (in a good way) of some of them. Also, I would’ve bet them some of them are naturally produced, like wiggling broken VGA cable.