[Feature][Frontend]: Deprecate `--enable-reasoning`

[gaocegege]

🚀 The feature, motivation and pitch

If --reasoning-parser is required when --enable-reasoning is set, why don't we just remove --enable-reasoning? We can assume the user wants to enable reasoning content by their specification of --reasoning-parser

from #12955 (comment)

We could deprecate it.

Alternatives

No response

Additional context

No response

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[vincent-4]

Necessary change or more of a feature proposal? @gaocegege will do this if you (or other project owners) can determine if you want this now.

[gaocegege]

You can see #12955 (comment)

I think it is not in high priority

[passerbythesun]

I have this doubt too. Since DeepSeek-R1 are already reasoning models, what does the flag --enable-reasoning acutally do to "enable" it? Maybe it's better to rename it if necessary.

[gaocegege]

We can deprecate it and keep --reasoning-parser deepseek_r1 to tell the vLLM engine to parse the reasoning_content.

[gaocegege]

Feel free to pick it up. I will work on this next week if no one is interested in it.

[chaunceyjiang]

Please assign it to me.

[gaocegege]

@chaunceyjiang Thanks! Please at me if you need a review. By the way, please make sure it works with both v0 and v1 engine.