tests/rng-op-key extrmely slow on the MinGW runners

I noticed it too, though these are not new tests, and the mingw runners were much faster than their current speed. We may be seeing the same MR so hopefully it is a glitch. If it persists we should check whether we can make this test faster (e.g., less iterations but still test what it is supposed to test). btw. yes it calls the random generator but it shouldn't be waiting for entropy... said that, I think there was a recent MR changing the API used by the random generator in windows...