Release: v2.0-beta Architecture, Visual Analytics UX, and CI/CD Automation

[FlorentLee]

🚀 Overview
This PR introduces the core upgrades for the spread.gl v2.0-beta release. It directly addresses the methodological and UI/UX critiques raised during the thesis review (specifically regarding the "black box" nature of backend filtering) and establishes a modern, fully automated DevOps pipeline for public deployment.

spread.gl now operates as a true Integrated Data Pipeline, decoupling heavy computation from a frictionless, highly interactive frontend.

📊 1. Interactive Visual Analytics & UX
We have shifted the filtering philosophy to empower the user, bringing the methodology strictly in line with modern visual analytics standards:

Dynamic Bayes Factor Filtering: The backend now computes and attaches Bayes Factors to all lineages without dropping data. Users can interactively explore the posterior distribution of the Dynamic Pathways using native Kepler.gl UI sliders.

Aggregated Network Pre-filtering: Set up a 0–200 Bayes Factor slider on the Setup page to intelligently prune macroscopic noise from the static network before rendering.

Frictionless Drag-and-Drop: Users can bypass the backend entirely by dropping local .geojson, .csv, or previously saved .json map states directly into the UI.

Synchronized 4D Timeline: Fully integrated Kepler's playback to simultaneously animate viral Trips, HPD contours, and Environmental Rasters.

🌍 2. Accessibility & Compliance
Tianditu (天地图) Integration: Added native support for China's official Tianditu basemaps, ensuring strict mapping compliance and vastly improving accessibility for Chinese academia and domestic publications.

⚙️ 3. DevOps, CI/CD, & Live Deployment
The project is now fully containerized and automated for both secure local execution and public demonstration:

Docker Hub CI/CD: Created a docker-publish.yml GitHub Action. Pushes to main now trigger a parallel matrix build, automatically compiling and pushing the v2.0-beta frontend and backend images to Docker Hub.

Hugging Face Live Demo: Engineered a unified Dockerfile utilizing an Nginx reverse-proxy to bundle the FastAPI backend and React frontend into a single container. The CI/CD pipeline now automatically syncs and deploys the live demo to Hugging Face Spaces on every release.

📝 4. Documentation
Complete README Overhaul: Updated the documentation to reflect the new 3-Step GUI nomenclature (Setup ➔ Workspace ➔ Map).

Added explicit, step-by-step pipeline walkthroughs for both the YFV Continuous and B.1.525 Discrete examples, highlighting the tool's scientific utility.

Reviewer Note: This PR contains no breaking changes to the core statistical algorithms (e.g., continuous_space_processor.py or discrete_space_processor.py), ensuring prior analytical validity is perfectly preserved.