docs: prioritize ADK in README feature highlights

- Move ADK above composition framework in 'What Eino provides' list
- Reorder 'Key Features' section to put ADK before Orchestration
- Update both English and Chinese README files consistently
- Adjust intro text to better position ADK as a primary feature

Change-Id: I27b41b8051f45900af63290f3c92711e2120b0fe

Author: shentongmartin

README.md

@@ -18,8 +18,8 @@ English | [中文](README.zh_CN.md)
 
 What Eino provides are:
 - a carefully curated list of **component** abstractions and implementations that can be easily reused and combined to build LLM applications
-- a powerful **composition** framework that does the heavy lifting of strong type checking, stream processing, concurrency management, aspect injection, option assignment, etc. for the user.
 - an **Agent Development Kit (ADK)** that provides high-level abstractions for building AI agents with multi-agent orchestration, human-in-the-loop interrupts, and prebuilt agent patterns.
+- a powerful **composition** framework that does the heavy lifting of strong type checking, stream processing, concurrency management, aspect injection, option assignment, etc. for the user.
 - a set of meticulously designed **API** that obsesses on simplicity and clarity.
 - an ever-growing collection of best practices in the form of bundled **flows** and **examples**.
 - a useful set of tools that covers the entire development cycle, from visualized development and debugging to online tracing and evaluation.
@@ -286,18 +286,9 @@ agent, _ := adk.NewChatModelAgent(ctx, &adk.ChatModelAgentConfig{
     - ReAct Agent, MultiQueryRetriever, Host MultiAgent, etc. They consist of multiple components and non-trivial business logic.
     - They are still transparent from the outside. A MultiQueryRetriever can be used anywhere that accepts a Retriever.
 
-## Powerful Orchestration
-
-- Data flows from Retriever / Document Loaders / ChatTemplate to ChatModel, then flows to Tools and parsed as Final Answer. This directed, controlled flow of data through multiple components can be implemented through **graph orchestration**.
-- Component instances are graph nodes, and edges are data flow channels.
-- Graph orchestration is powerful and flexible enough to implement complex business logic:
-  - type checking, stream processing, concurrency management, aspect injection and option assignment are handled by the framework.
-  - branch out execution at runtime, read and write global state, or do field level data mapping using workflow(currently in alpha stage).
-  - **Aspects (Callbacks)** handle cross-cutting concerns such as logging, tracing, and metrics. Five aspects are supported: OnStart, OnEnd, OnError, OnStartWithStreamInput, OnEndWithStreamOutput. Custom callback handlers can be added during graph run via options.
-
 ## Agent Development Kit (ADK)
 
-While graph orchestration gives you fine-grained control, the **ADK** package provides higher-level abstractions optimized for building AI agents:
+The **ADK** package provides high-level abstractions optimized for building AI agents:
 
 - **ChatModelAgent**: A ReAct-style agent that handles tool calling, conversation state, and the reasoning loop automatically.
 - **Multi-Agent with Context Engineering**: Build hierarchical agent systems where conversation history is automatically managed across agent transfers and agent-as-tool invocations, enabling seamless context sharing between specialized agents.
@@ -306,6 +297,22 @@ While graph orchestration gives you fine-grained control, the **ADK** package pr
 - **Prebuilt Patterns**: Ready-to-use implementations including Deep Agent (task orchestration), Supervisor (hierarchical coordination), and Plan-Execute-Replan.
 - **Agent Middlewares**: Extensible middleware system for adding tools (filesystem operations) and managing context (token reduction).
 
+## Powerful Orchestration
+
+For fine-grained control, Eino provides **graph orchestration** where data flows from Retriever / Document Loaders / ChatTemplate to ChatModel, then flows to Tools and parsed as Final Answer.
+
+- Component instances are graph nodes, and edges are data flow channels.
+- Graph orchestration is powerful and flexible enough to implement complex business logic:
+  - type checking, stream processing, concurrency management, aspect injection and option assignment are handled by the framework.
+  - branch out execution at runtime, read and write global state, or do field level data mapping using workflow.
+
+## Aspects (Callbacks)
+
+**Aspects** handle cross-cutting concerns such as logging, tracing, and metrics. They can be applied to components directly, orchestrated graphs, or ADK agents.
+
+- Five aspect types are supported: OnStart, OnEnd, OnError, OnStartWithStreamInput, OnEndWithStreamOutput.
+- Custom callback handlers can be added at runtime via options.
+
 ## Complete Stream Processing
 
 - Stream processing is important because ChatModel outputs chunks of messages in real time as it generates them. It's especially important with orchestration because more components need to handle streaming data.

README.zh_CN.md

@@ -18,8 +18,8 @@
 
 Eino 提供的价值如下：
 - 精心整理的一系列 **组件（component）** 抽象与实现，可轻松复用与组合，用于构建 LLM 应用。
-- 强大的 **编排（orchestration）** 框架，为用户承担繁重的类型检查、流式处理、并发管理、切面注入、选项赋值等工作。
 - **智能体开发套件（ADK）**，提供构建 AI 智能体的高级抽象，支持多智能体编排、人机协作中断机制以及预置的智能体模式。
+- 强大的 **编排（orchestration）** 框架，为用户承担繁重的类型检查、流式处理、并发管理、切面注入、选项赋值等工作。
 - 一套精心设计、注重简洁明了的 **API**。
 - 以集成 **流程（flow）** 和 **示例（example）** 形式不断扩充的最佳实践集合。
 - 一套实用 **工具（DevOps tools）**，涵盖从可视化开发与调试到在线追踪与评估的整个开发生命周期。
@@ -285,18 +285,9 @@ agent, _ := adk.NewChatModelAgent(ctx, &adk.ChatModelAgentConfig{
     - ReAct Agent、MultiQueryRetriever、Host MultiAgent 等。它们由多个组件和复杂的业务逻辑构成。
     - 从外部看，它们的实现细节依然透明。例如在任何接受 Retriever 的地方，都可以使用 MultiQueryRetriever。
 
-## 强大的编排 (Graph/Chain/Workflow)
-
-- 数据从 Retriever / Document Loader / ChatTemplate 流向 ChatModel，接着流向 Tool ，并被解析为最终答案。这种通过多个组件的有向、可控的数据流，可以通过**图编排**来实现。
-- 组件实例是图的 **节点（Node）** ，而 **边（Edge）** 则是数据流通道。
-- 图编排功能强大且足够灵活，能够实现复杂的业务逻辑：
-    - **类型检查、流处理、并发管理、切面注入和选项分配**都由框架处理。
-    - 在运行时进行**分支（Branch）**执行、读写全局**状态（State）**，或者使用工作流进行字段级别的数据映射。
-    - **切面（Callbacks）** 处理日志记录、追踪、指标统计等横切关注点。支持五种切面：OnStart、OnEnd、OnError、OnStartWithStreamInput、OnEndWithStreamOutput。可通过 Option 在图运行时添加自定义回调处理程序。
-
 ## 智能体开发套件（ADK）
 
-图编排提供细粒度控制，而 **ADK** 包则提供了针对构建 AI 智能体优化的更高级抽象：
+**ADK** 包提供了针对构建 AI 智能体优化的高级抽象：
 
 - **ChatModelAgent**：ReAct 风格的智能体，自动处理工具调用、对话状态和推理循环。
 - **多智能体与上下文工程**：构建层级化智能体系统，对话历史在智能体转移和智能体作为工具调用时自动管理，实现专业智能体间的无缝上下文共享。
@@ -305,6 +296,22 @@ agent, _ := adk.NewChatModelAgent(ctx, &adk.ChatModelAgentConfig{
 - **预置模式**：开箱即用的实现，包括 Deep Agent（任务编排）、Supervisor（层级协调）和 Plan-Execute-Replan。
 - **智能体中间件**：可扩展的中间件系统，用于添加工具（文件系统操作）和管理上下文（token 缩减）。
 
+## 强大的编排 (Graph/Chain/Workflow)
+
+如需细粒度控制，Eino 提供**图编排**能力，数据从 Retriever / Document Loader / ChatTemplate 流向 ChatModel，接着流向 Tool ，并被解析为最终答案。
+
+- 组件实例是图的 **节点（Node）** ，而 **边（Edge）** 则是数据流通道。
+- 图编排功能强大且足够灵活，能够实现复杂的业务逻辑：
+    - **类型检查、流处理、并发管理、切面注入和选项分配**都由框架处理。
+    - 在运行时进行**分支（Branch）**执行、读写全局**状态（State）**，或者使用工作流进行字段级别的数据映射。
+
+## 切面（Callbacks）
+
+**切面**处理日志记录、追踪、指标统计等横切关注点。切面可以直接应用于组件、编排图或 ADK 智能体。
+
+- 支持五种切面类型：OnStart、OnEnd、OnError、OnStartWithStreamInput、OnEndWithStreamOutput。
+- 可通过 Option 在运行时添加自定义回调处理程序。
+
 ## 完整的流式处理能力
 
 - 流式处理（Stream Processing）很重要，因为 ChatModel 在生成消息时会实时输出消息块。在编排场景下会尤为重要，因为更多的组件需要处理流式数据。