refactor: 训练循环改为两重迭代并统一参数命名

- 训练循环从三重(epoch→batched→batch)改为二重(epoch→batch)
- batch_size → batch_per_device, accumulation_steps → grad_accum_steps
- scheduler 移入 step block 对齐 optimizer 更新步
- GradientClippingCallback 改用 on_step_begin 避免零梯度裁剪
- 移除 _train_impl 误导性的 -> Checkpoint 标注
- total_steps 修除为向下取整并精简为一行
- warmup_steps 改为 warmup_ratio (默认0.05)

astrai/config/train_config.py

@@ -20,8 +20,10 @@ class TrainConfig:
         default=None, metadata={"help": "Scheduler factory for training."}
     )
     n_epoch: int = field(default=1, metadata={"help": "Number of epochs for training."})
-    batch_size: int = field(default=4, metadata={"help": "Batch size for training."})
-    accumulation_steps: int = field(
+    batch_per_device: int = field(
+        default=4, metadata={"help": "Batch size per device."}
+    )
+    grad_accum_steps: int = field(
         default=1, metadata={"help": "Number of iterations between steps."}
     )
     max_grad_norm: float = field(