docs: 同步 architecture/inference/training 文档至实际代码，CLI 补充 fsdp 选项

- 修正 ProtocolHandler 架构：concrete + ResponseBuilder(ABC) 策略模式
- 修正训练循环 scheduler.step() 在 sync_gradients 块内
- 修正组合/聚合关系：注入组件改为 o--，删除不持有引用的关联
- --parallel_mode CLI choices 加入 fsdp
- nprocs > 1 且 parallel_mode=none 时 raise error

assets/docs/dataflow.md

@@ -33,14 +33,14 @@ Both support shared memory via `.share_memory_()`.
 ## Dataset Architecture
 
 ```
-DatasetFactory.load(train_type, path, window_size, stride)
+DatasetFactory.load(train_type, path, window_size, stride, storage_type, tokenizer)
   → StorageFactory.create(detect_format(path))
     → MultiSegmentFetcher(BaseSegmentFetcher per key)
       → BaseDataset.__getitem__(idx)
         → sliding window [begin, end) via get_index(idx)
 ```
 
-`window_size` = max input length, `stride` = step between consecutive samples.
+`window_size` = max input length, `stride` = step between consecutive samples (defaults to `window_size`).
 
 ## Sampler