docs: add mathematical formulas, quality retention algorithm, and individual prompt evaluation examples

fix: populate live analytics dashboard with real WildChat 10k benchmark metrics

feat: integrate allenai/WildChat-4.8M as flagship real dataset & update 10k benchmark results

feat: add HF real workload datasets (10k), streaming benchmark results, and BYOK budget tests

feat: integrate 9 academic routing policies as explicit selectable gateway strategies and add benchmarking harness

feat: implement server-side FrugalGPT cascade router, prompt adaptation, and offline Pareto sweeps

feat: implement learned router and reproducible benchmark harness