> Plus there's zero direct evidence for their colours

This is no longer true! Starting with Sinosauropteryx in 2010, paleontologists have identified what they believe to be fossilized melanin-containing organelles. These organelles, called melanosomes, have different shapes depending on which color they produce, and those shapes are preserved well enough to be visible under an electron microscope.

Amazing, thanks for pointing it out. In the meantime, there's been some rejigging of the classification so it's this related genus where they've found the melanosomes

https://en.wikipedia.org/wiki/Huadanosaurus

Isn't a rather good deal of color from feathers a result of "structural color", rather than pigmentation? I'd be curious if fossilized feathers could ever, in theory, preserve enough microscopic detail to guess at that.