[Bug]: Vertex AI Prompt Optimizer produces no optimization results for Gemini 3.x target models (gemini-3.0-flash, gemini-3.1-flash-lite)

### File Name

get_started_with_vertex_ai_prompt_optimizer.ipynb , vapo_lib.py

### What happened?

### **Description**

I am testing Vertex AI Prompt Optimizer with different Gemini target models.

The optimization workflow works correctly when using Gemini 2.5 models as the target model. Optimized prompts and output artifacts are generated as expected.

However, when I use Gemini 3.x models (for example gemini-3.0-flash or gemini-3.1-flash-lite) as the target model, the Prompt Optimizer job is successfully created and runs, but no optimization results are produced.

### Relevant log output

```shell
**Observed Behavior**
Prompt Optimizer job is submitted successfully.
Output bucket is created/populated with initial artifacts.
No optimized prompt results are generated.
No explicit error is shown in the UI.
The same workflow works with Gemini 2.5 target models.
Expected Behavior

Prompt Optimizer should either:

Generate optimization results for Gemini 3.x target models, or
Return a clear validation/error message if Gemini 3.x target models are not currently supported.
```

### Code of Conduct

- [x] I agree to follow this project's Code of Conduct