feat: add Markdown-to-image for notification channels (Fixes #289) (#315)

* feat: add Markdown-to-image for notification channels (Fixes #289)

- Add MARKDOWN_TO_IMAGE_CHANNELS config for telegram, wechat, custom, email
- New src/md2img.py: markdown_to_image() via imgkit/wkhtmltoimage
- Extract markdown_to_html_document to src/formatters.py for reuse
- Implement image sending: Telegram sendPhoto, WeChat base64+md5,
  Discord webhook multipart, email inline attachment
- Add imgkit to requirements, wkhtmltopdf to Dockerfile
- Graceful fallback to text when imgkit/wkhtmltopdf unavailable

* fix(docker): pin to bookworm for wkhtmltopdf, add runtime deps

wkhtmltopdf was removed from Debian testing (2025). Pin base image to
python:3.11-slim-bookworm. Add fontconfig, libjpeg62-turbo, libxrender1,
libxext6 for headless wkhtmltopdf.

* docs(md2img): add security note on command injection and XSS

* refactor(md2img): extract _should_use_image_for_channel, document Fallback rules

* docs: add Markdown-to-image dependency install guide (imgkit, wkhtmltopdf)

requirements.txt

@@ -35,6 +35,7 @@ google-search-results>=2.4.0  # SerpAPI（每月 100 次免费）
 # 网络请求
 requests>=2.31.0            # HTTP 请求
 markdown2>=2.4.0            # Markdown 转 HTML
+imgkit>=1.2.0              # Markdown 转图片（需安装 wkhtmltopdf）
 fake-useragent>=1.4.0       # 随机 User-Agent 防封禁
 httpx[socks]                # HTTP 客户端 + SOCKS 代理支持（OpenAI 可选依赖）
 dingtalk-stream >= 0.24.3    # 钉钉 Stream SDK