fix: specify utf-8 encoding in subprocess.run() to prevent UnicodeDecodeError on Windows (fixes #641)

On Windows, subprocess.run() with text=True defaults to the system locale
encoding (e.g. GBK), which cannot decode UTF-8 subprocess output and raises
UnicodeDecodeError during database initialization.

Explicitly set encoding='utf-8' in both subprocess.run() calls that use
text=True in MindSpider/main.py:
- initialize_database(): runs init_database.py and captures its output
- _install_mediacrawler_dependencies(): runs pip/uv install commands