NvEnc Limit Stop Testing for device

[nathangur]

Since NVIDIA limits the ammount of concurrent NvEnc Streams on their Consumer hardware, the script needs to be abled to understand "nvenc_limit" (or smth like that) as a failure reason.

However the way up and downscaling (of the worker ammount per run) currently works in a way, where we upscale by speed and then downscale until its working again.

For runs on NvEnc limited hardware that would mean useless scaledowns, only to stay at "8" workers.

Therefore the script should be abled to detect after the first GPU test, that the Device is NvEnc limited and will not try to go past that limit on the other runs.