[Refactor][Frontend] Keep all logic about reasoning into one class (#14428)

Signed-off-by: Ce Gao <cegao@tensorchord.ai>

vllm/entrypoints/openai/serving_chat.py

@@ -23,8 +23,6 @@ from vllm.entrypoints.openai.protocol import (
     ChatCompletionStreamResponse, ChatMessage, DeltaFunctionCall, DeltaMessage,
     DeltaToolCall, ErrorResponse, FunctionCall, PromptTokenUsageInfo,
     RequestResponseMetadata, ToolCall, UsageInfo)
-from vllm.entrypoints.openai.reasoning_parsers import (ReasoningParser,
-                                                       ReasoningParserManager)
 from vllm.entrypoints.openai.serving_engine import (OpenAIServing,
                                                     clamp_prompt_logprobs)
 from vllm.entrypoints.openai.serving_models import OpenAIServingModels
@@ -33,6 +31,7 @@ from vllm.entrypoints.openai.tool_parsers.mistral_tool_parser import (
     MistralToolCall)
 from vllm.logger import init_logger
 from vllm.outputs import CompletionOutput, RequestOutput
+from vllm.reasoning import ReasoningParser, ReasoningParserManager
 from vllm.sampling_params import BeamSearchParams, SamplingParams
 from vllm.sequence import Logprob
 from vllm.transformers_utils.tokenizer import AnyTokenizer, MistralTokenizer