fix(vllm): tell the reasoning parser whether thinking was enabled

[pos-ei-don]

Description

vLLM's engine-based reasoning parsers derive their initial state from the chat template
kwargs. Qwen3Parser:

chat_kwargs = kwargs.get("chat_template_kwargs", {}) or {}
self.thinking_enabled = chat_kwargs.get("enable_thinking", True)

Constructed as ReasoningParser(tokenizer) the flag defaults to True, so the parser starts
in the REASONING state. A completion produced with thinking disabled contains no tags at all,
and every reasoning parser shape then reports the whole answer as reasoning:

• engine-based parsers classify it by initial state;
• BaseThinkingReasoningParser hits its documented "may not generate start token" fallback
  and returns (model_output, None).

Either way content = c if c is not None else generated_text turns that into a duplicate: a
Qwen3 model answering 391 with thinking off comes back as reasoning_content="391" and
content="391".

Measured against Qwen3.5-MoE on vLLM 0.28, non-streaming:

before   thinking on   reasoning=202   content="391"
         thinking off  reasoning="391" content="391"   <- duplicated
after    thinking on   reasoning=192   content="391"
         thinking off  reasoning=""    content="391"

The fix forwards the kwargs the prompt was rendered with, which is what vLLM's own OpenAI
server does; parsers that do not accept the argument keep the plain constructor.

Notes for Reviewers

The parser class matters here. On vLLM 0.28 the qwen3 name resolves to
Qwen3ParserReasoningAdapter, an engine-based adapter with no start_token — not to
BaseThinkingReasoningParser. A fix written against the latter is inert against the shipped
parser, which is why this one sets the state through the constructor rather than around it.

_split_reasoning() covers the older parser shape, which has no initial state to set. It
only reclassifies when the parser exposes a start/end token pair and neither the
completion nor the prompt ever opened a reasoning block. Truncated reasoning (block open, end
token never arrived) stays reasoning, and parsers without that token pair are left untouched.

Worth noting for anyone reading this alongside reasoning_effort: vLLM forwards the value
into template kwargs but filters it out when the template does not declare it. For a template
that only knows enable_thinking, thinking is binary and reasoning_effort has no effect
beyond on/off — this PR does not change that.

Signed commits

• Yes, I signed my commits.
• Documentation updated (docs/content/) for user-facing changes, or not applicable

[localai-org-maint-bot]

Good to merge from my review. The parser now receives the same template kwargs used to render the prompt, while the fallback preserves older parser compatibility; the split logic also distinguishes plain answers from truncated reasoning. Regression coverage, Python syntax checks, and diff checks look clean. @mudler