OCR扫描pom后结论中显示Java代码，扫描文件和分析对不上

### OpenCodeReview Version

open-code-review v1.8.8 (6dc3eb0)

### Operating System

Linux (x86_64)

### Installation Method

npm (global)

### LLM Provider

Other OpenAI-compatible endpoint

### Bug Description

{
"path": "pom.xml",
"content": "`toString()` 方法不应使用 `System.out.println()` 进行输出，而应仅通过 `return` 返回对象的字符串表示。当前实现会导致每次调用 `toString()` 时产生大量标准输出，且调用方（如日志框架、调试器等）无法获取到任何有意义的返回值（虽然 return 也有字符串，但大量 println 会造成副作用和性能问题）。应移除所有 `System.out.println` 调用，只保留 `return` 语句。",
"existing_code": " public String toString() {\n System.out.println(\"id=\" + id );\n System.out.println(\"resourceId=\" + resourceId);\n System.out.println(\"name=\" + name);\n System.out.println(\"alias=\" + alias);",
"start_line": 0,
"end_line": 0,
"category": "bug",
"severity": "high"
},

### Steps to Reproduce

ocr review \
--from "origin/test" \
--to "origin/dev" \
--format json \
--concurrency 4 \
--audience agent --format json > ocr-result.json

### Expected Behavior

ocr扫描报告能准确匹配

### Logs / Error Output

```shell

```

### Additional Context

_No response_

Reproduce the issue with the shown `ocr review` command using the `origin/test` and `origin/dev` revisions, then inspect `ocr-result.json`. Compare the reported `path`, `content`, `existing_code`, and line fields for `pom.xml` with the scanned source. Done means the report identifies the actual scanned file and matching Java code.

Written by the indexing model from the issue text.