Allow the API to only return non-reasoning output from reasoning models.

Creating a flag for API users that strips reasoning output from the LLM-generated answer would be a great addition to Venice.

It would allow users to use some of the interesting models like DeepSeek or Qwen on libraries that do not intrinsically support stripping the reasoning output.

A lot of newer libraries do not natively differentiate between reasoning and non-reasoning models which are exactly the kind of libraries that might pique the interest of some of the more experimentally minded Venice AI builders.