fix : 并行训练 state_dict 收集与训练/推理并发缺陷

- FSDPExecutor: unwrap_model 返回全量 state_dict (state_dict_type FULL)；use_orig_params=True
- DDPExecutor/BaseExecutor: unwrap_model 统一返回 model.module.state_dict() / model.state_dict()
- CheckpointCallback: 走 executor.unwrap_model 拿完整 state_dict
- strategy.py: 移除 FSDP/DDp 依赖；create_ref_model(model_fn, state_dict) 纯函数
- TrainContextBuilder: 传递 model_fn + executor 到 strategy
- GRPOStrategy.sync_ref_model: 通过 executor.unwrap_model 获取完整权重
- TaskManager.wait_for_tasks: 锁内检查队列，消除 clear/set 竞态
- ProtocolHandler: stop token 不再计入 completion_tokens（流式/非流式）

astrai/inference/api/protocol.py

@@ -138,13 +138,13 @@ class ProtocolHandler:
             yielded = ""
             matched = None
             async for token in agen:
-                ctx.completion_tokens += 1
                 body += token
 
                 matched = checker.check(body)
                 if matched:
                     break
 
+                ctx.completion_tokens += 1
                 yield self.builder.format_chunk(token)
                 yielded += token
 
@@ -168,7 +168,6 @@ class ProtocolHandler:
         matched = None
 
         async for token in agen:
-            ctx.completion_tokens += 1
             chunks.append(token)
             body += token
 
@@ -176,6 +175,8 @@ class ProtocolHandler:
             if matched:
                 break
 
+            ctx.completion_tokens += 1
+
         content = "".join(chunks)
         stop = StopInfo(matched=matched, body=body)
         return self.builder.format_response(ctx, content, stop)