docs: refine bilingual README and add recognition demos

[SakuraMathcraft]

Summary

• Refine both English and Chinese READMEs around product capabilities, download/setup, demos, and links to detailed documentation.
• Highlight downloads and the Chinese PDF manual, reduce badges to four, merge overlapping features, and replace implementation-heavy platform/API prose with user-facing requirements and security essentials.
• Add three original-speed recognition demos as 960x540, approximately 30 fps GIFs. Show screenshot recognition by default; place image/PDF demos in collapsible sections.
• Align bilingual sections and translate English sponsorship labels, preserving all supporters and community links.

Validation

• Checked all local Markdown/image links and balanced details/summary tags in both READMEs.
• Verified GIF dimensions, frame counts, and duration; reviewed extracted demo frames during preparation.
• git diff --check passed; no application code or runtime behavior changes.
• Full pytest/installer builds not run for this documentation-only change.

Asset tradeoff

The three GIFs total approximately 41 MB, preserving the requested 30 fps animation. Collapsing sections reduces visible page length but does not guarantee reduced network transfer. Original MP4 files remain unchanged outside the repository.

[SakuraMathcraft]

MP4 versions of the README demos, replacing the large GIFs. Encoded as H.264 at 1920x1080, 60 fps, original playback speed, without audio.

Screenshot Recognition

screenshot-recognition.mp4

Image Recognition

image-recognition.mp4

PDF Recognition

pdf-recognition.mp4