We are working on a paper right now with something similar to your hypothesis "Its pretty reasonable for a metal ion to be trapped in all those sulfurs, then oxidized from a colorless form to colored one by light." So I'm sympathetic...except that the degradation seems to happen even when the starting bithiophene is real, real pure. I suppose that traces of nickel (it's Kumada coupling that's used, btw, or occasionally Suzuki...Grignard the 2-bromothiophene, then drip the organometallic into a Ni(II) catalyst with more 2-bromothiophene) could be retained even after re-recrystallizing it, but it seems just a little unlikely. Yes, easy to check with a Ni method, maybe the spring instrumental class can do a flame AA on my 2,2'-bi.
Another thought: the oxidized stuff, 2,2-dioxides of thiophenes, can act as Diels-Alder dienes. I wonder if that kind of crosslinking can give something green or blue. Also far-fetched.