fix: support both old and new Bailian Rerank API response formats

[he-yufeng]

Summary

commit 971bcba 要求 qwen3-rerank 使用新版 compatible API (/compatible-api/v1/reranks)，但没有更新响应解析逻辑。新 API 返回 data.results，而解析器只检查旧路径 data.output.results，导致 qwen3-rerank 始终返回空结果。

改为同时兼容两种格式：优先检查 output.results（旧 API），回退到 results（新 API）。

Changes

• bailian_rerank_source.py: _parse_results() 兼容新旧两种响应格式

Test

1. 配置阿里云百炼 qwen3-rerank 模型
2. 使用新版 API https://dashscope.aliyuncs.com/compatible-api/v1/reranks
3. 点击测试，验证能正确返回重排序结果

Fixes #7161

Summary by Sourcery

Bug Fixes:

• Ensure Bailian Rerank results are correctly parsed when using the new compatible API response structure that returns data.results instead of data.output.results.

[sourcery-ai]

Hey - I've found 1 issue, and left some high level feedback:

• Using or to fall back from data.get('output', {}).get('results') to data.get('results', []) conflates a legitimately empty list with a missing field; consider explicitly checking for is None so that an intentionally empty output.results is not silently overridden by results when both are present.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- Using `or` to fall back from `data.get('output', {}).get('results')` to `data.get('results', [])` conflates a legitimately empty list with a missing field; consider explicitly checking for `is None` so that an intentionally empty `output.results` is not silently overridden by `results` when both are present.

## Individual Comments

### Comment 1
<location path="astrbot/core/provider/sources/bailian_rerank_source.py" line_range="145-146" />
<code_context>
             )

-        results = data.get("output", {}).get("results", [])
+        # 兼容旧版 API (output.results) 和新版 compatible API (results)
+        results = data.get("output", {}).get("results") or data.get("results", [])
         if not results:
             logger.warning(f"百炼 Rerank 返回空结果: {data}")
</code_context>
<issue_to_address>
**question:** Consider whether falling back on `or` should treat an empty legacy `output.results` as "missing".

Using `or` here means an empty `output.results` will be treated the same as `None` and replaced by `data["results"]`, so the warning/early return is no longer driven by the legacy field being empty. If the legacy API can legitimately return `[]` and that should remain authoritative when `results` is also present, consider an explicit `is not None` check instead:

```python
legacy_results = data.get("output", {}).get("results")
if legacy_results is not None:
    results = legacy_results
else:
    results = data.get("results", [])
```

If you do want `results` to win whenever `output.results` is falsy, the current code is fine.
</issue_to_address>

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

[gemini-code-assist]

Code Review

This pull request updates the _parse_results method in the Bailian rerank source to ensure compatibility with both legacy and new API response formats. The feedback identifies a potential AttributeError if the 'output' field is explicitly null and provides a suggestion to improve the robustness of the data extraction logic.