The bigger problem is VRM wear, you never know when it will fail. Said TIM drying out affects the VRM components the most. One failed capacitor or resistor and the whole GPU can get fried (seen it happen many times).
Is there a good way to spot this sort of thing from a distance when evaluating stuff off eg eBay?
SMD elements can fail either open or closed/shorted, the latter is more likely to cause a worse failure but it's not exclusive. Due to protection circuits elsewhere they can fail without visible marks so it's hard to diagnose.
Thermal pads dry out because of heat first, their quality second, with load coming in last imo. So the fact that it's been used for mining alone doesn't mean much.
The conditions under which a card was used and the time matter more - if only sellers told you the average temperature, case type, ambient temperatures, whether the cards were undervolted/overclocked and how long they actually ran...