release: v2.1.0 — corpus-trained humanization engine

- Corpus-trained engine from 10,000 Q1 papers (11 domains, 2018-2025)
- Dynamic detection thresholds calibrated against real human writing
- Expanded collocation database (150+ AI phrase replacements)
- Style-aware prompt builder with corpus-calibrated statistical targets
- Fixed LinkedIn URL, workflow triggers, client bundle crash, markdown lint
- Updated README, SECURITY, ROADMAP, SUPPORT, docs, LICENSE, package.json

Major SEO + social + security upgrade

- Massive README rewrite with badges, comparison tables, full docs
- SECURITY.md with vulnerability reporting policy
- FUNDING.yml
- SEO: meta tags, Open Graph, Twitter Card, JSON-LD structured data
- robots.txt + sitemap.xml for GitHub Pages
- Social links: Twitter (@rudra496), GitHub, portfolio site
- Author info: Rudra Sarker, SUST Bangladesh
- Footer upgraded with full social links
- OG image placeholder