feat: implement self-healing online model retraining loop on statistical data drift detection

feat: upgrade fraud detection to XGBoost & SHAP local explanations with premium dashboard

build: completely purge emergent / emergentintegrations references and packages

build(frontend): install devDependencies explicitly during docker production build stage to include craco

build(frontend): purge eslint block from craco config and secure docker build memory resources

build(frontend): remove private visual-edits package and bypass webpack eslint on production build

build(deps): remove unused jq and librt from requirements to prevent Docker build failure

perf(search): offload Cross-Encoder reranking model inference to thread pools asynchronously

fix(backend): remove undefined client reference in FastAPI shutdown handler

fix: format list embeddings as PostgreSQL vector strings for asyncpg compatibility

feat: migrate and consolidate all operational logging, structured tracing, and metrics to Supabase PostgreSQL, removing MongoDB Atlas dependency

feat: make backend resilient against PostgreSQL pool initialization crashes on startup

chore: align date-fns version to resolve peer dependency mismatches

feat: complete NeuralVault 3.0 hardening including AST parameterization fix, KS data drift, Cross-Encoder, and Docker Compose

docs: prepend Hugging Face Spaces YAML metadata card

feat: NeuralVault 2.0 - Complete Free-Tier Groq & ML Explainability Architecture