feat: Add SaaS models, pricing infrastructure, and benchmark support

• Add SaaS API models section (GPT-5, Claude, Gemini, GPT-4o)
• Add benchmark support (SWE-Bench, HumanEval, BFCL)
• Add model performance chart with dynamic benchmark descriptions
• Add pricing infrastructure with manual update workflow
• Fix table horizontal scrolling in collapsible sections
• Move benchmark explanations above model specifications
• Add .env.example for environment variables
• Update GPT-5 mini and GPT-OSS scores
•  Family/Provider column with color-coded badges

Technical improvements:

• Restructured collapsible cards for proper table scrolling
• Made all major sections collapsible
• Added dynamic descriptions that update when switching benchmarks
• Documented future automation options for pricing updates

Files changed:

• components/model-performance-chart.tsx: Enhanced chart with benchmarks
• components/reference-documentation.tsx: Added SaaS section, fixed tables
• lib/data.ts: Added SaaS models and benchmark scores
• lib/pricing-history.json: GPU pricing data structure
• .env.example: Environment variable template