Fix: Explicit UTF-8 encoding for cross-platform file reading

[AQGitHbu]

Problem

On Windows (especially Chinese/Asian locales) and some other systems, the default file encoding is not UTF-8. This causes UnicodeDecodeError when reading session files that contain non-ASCII characters.

Solution

Explicitly specify encoding="utf-8" in all open() and read_text() calls when reading JSON files.

Changes

• Line 82: Add encoding="utf-8" to read_text() in _extract_project_path_from_sessions()
• Line 124: Add encoding="utf-8" to open() in _iter_jsonl()
• Line 266: Add encoding="utf-8" to read_text() in _load_kimi_work_dirs()

Benefits

1. Cross-platform compatibility (Windows, Linux, macOS)
2. Consistent behavior regardless of system locale
3. Properly handles session files with non-ASCII content

[woct0rdho]

There are more places that we need to specify encoding="utf-8". I think an easier way is to force it run in Python UTF-8 mode.

You can see other fixes I did to support Windows at woct0rdho/dataclaw@anonymizer...windows