fix: 修复文档多处不准确 + inference scheduler 越界 bug + SchedulerCallback 回调阶段修正

文档 (6 个文件):
- design.md: 15+ 处修正 — persistent_key_values→paged_cache,
  MLA 字段重写, Server/ParallelSetup 不存在类移除,
  关系箭头方向修复, SchedulerCallback 阶段修正等
- dataflow.md: 重写数据流图和描述, 修复训练回调顺序、
  数据键名、MLA 归属、MetricTracker 等错误
- introduction.md: 层数 32→24, MLP 图双 Linear 修正,
  默认值/响应字段/health 端点修复
- params.md: 补充 grpo 及 4 个 GRPO 参数
- README.md / README-zh-CN.md: generate.py 补全必需参数,
  删除重复注释, HuggingFace 声明修正

代码 (2 个文件):
- scheduler.py: n_pages 池加 page_size 余量防止越界;
  decode 前预分配页
- train_callback.py: SchedulerCallback 从 on_step_end 改
  回 on_batch_end (按 batch 步进学习率)

astrai/inference/scheduler.py

@@ -105,7 +105,9 @@ class InferenceScheduler:
         n_kv_heads = config.n_kv_heads
         head_dim = config.dim // config.n_heads
         n_layers = config.n_layers
-        n_pages = (max_batch_size * self.max_seq_len + page_size - 1) // page_size
+        n_pages = (
+            max_batch_size * (self.max_seq_len + page_size) + page_size - 1
+        ) // page_size
 
         self.page_cache = PagedCache(
             n_layers,
@@ -279,6 +281,9 @@ class InferenceScheduler:
         tasks = sorted(tasks, key=lambda t: t.task_id)
         batch_sz = len(tasks)
 
+        for t in tasks:
+            self._maybe_alloc_page(t, start_pos)
+
         input_ids = torch.zeros(batch_sz, dtype=torch.long, device=self.device)
         for i, t in enumerate(tasks):
             input_ids[i] = t.output_ids[-1] if t.output_ids else t.prompt_ids[-1]