Function AgentInfo.load_from_agent error when work on multiple agent

#### Environment details

- OS type and version:
- Python version: `3.10`
- pip version: `25.3`
- `google-cloud-aiplatform` version: `1.126.1`

#### Steps to reproduce

I'm following this guideline to evaluate the agent based on adk framework
https://docs.cloud.google.com/agent-builder/agent-engine/evaluate#run-agent-evaluation
But I get the error when implement with multiple agent system

```
File "/Users/Emotiv/ELLIA/.venv/lib/python3.10/site-packages/vertexai/_genai/types/evals.py", line 111, in load_from_agent
instruction=agent.instruction,
File "/Users/Emotiv/ELLIA/.venv/lib/python3.10/site-packages/pydantic/main.py", line 1026, in __getattr__
raise AttributeError(f'{type(self).__name__!r} object has no attribute {item!r}')
AttributeError: 'SequentialAgent' object has no attribute 'instruction'
```
I think this api does not support for multiple agent. Just for single agent at the moment.

从 vertexai/_genai/types/evals.py 中的 AgentInfo.load_from_agent 开始，使用 SequentialAgent 重现文档化的 agent 评估流程。将那里访问的字段与 traceback 中显示的多 agent 对象进行比较；当评估流程能够处理多 agent 情况且不引发 AttributeError 时，即表示完成。

由索引模型根据 Issue 内容生成。