Nice! It sounds like you took extra good care of your iMac as I have done with my 1.25GHz one. That's definitely a lucky break that yours cooperated and released its motherboard from the sinks without the issues like the one from my friend. I was thinking that there was so much dust in that one that maybe it restricted airflow and ended up overheating and baking/curing the thermal paste to a hardened state with much higher tensile properties in adhesion and cohesion. I'm really glad I took my time and did not get impatient and try to force it.
I'm not sure about the cold solder theory, if I have to open them up again I will certainly look closely at all of the joints. I wonder though, why now and why both of them? I have heard of cold solder joints as faults in new systems but not developing over extended time periods. They are both 10-11 years old and the startup issues only started during the last few years with the older one seemingly a bit worse in a manner commensurate with its greater age. Since the power supplies are identical for the 1.0 and 1.25 GHz systems it sounds like a failure mode that is likely common between the 2 systems.
I will certainly learn more tomorrow morning when they get the cold room cold boot test and will post back with the results.
Brian