I think one thing to note is that dGPU problems are not as slowly degrading as you seem to allude towards. It's more of a switch. On =working. Off =broken. Just because it's working now doesn't mean it cant break any second.
The threshold to when then GPUs breaks, if it does, it might not, is very slight and quick. You could run a defective GPU to 99% of its death threshold and it could run forever. That one time for whatever reason it hits 100% its toast.
Also You can watch with a heat monitor internal temperatures, but what happens if you are GPU encode something or rendering something In the GPU. As soon as you see the heat jump up your gonna shut your computer down? Harder with a laptop cause of the battery. But doesn't that seem impossible? If the GPU hit that 100% braking point it's toast before you can do anything to to stop it.
I think my point is you can't say a GPU is healthy just because it's never been fried before.
I think. If you wanted to get serious. I Would see if Apple replaced the 17" close to the serial numbers you want to buy.
noted ! 🙂 thanks for replying back. 🙂 and i think i should allow myself to be optimistic for 2 seconds: what if the machine indeeds perform ok for years and here i am worrying if i should return it straight away. I guess i am tired of making assumptions after all these days.