HTTP timeout is too low

[yocontra]

https://github.com/runpod-workers/cog-worker/blob/main/src/handler.py#L49

600 seems low, especially for some jobs we are running we hit this timeout pretty often. Can we up this to what the actual runpod limit is, or remove it (Infinity) and let runpod itself manage killing a job that times out?

[nom]

+1 - ran into this unexpected timeouts and realized it was this line. Runpod offers an execution timeout option during the serverless endpoint creation which accomplishes the same thing. Would be better to let it be managed explicitly there.