## TurboQuant Integration (ml/embeddings/)
- Implemented TurboQuantKVCache: 3-bit PolarQuant angle quantization
  with QJL 1-bit sign correction using pure NumPy (no torch dependency)
- Implemented TurboQuantEmbeddings: LangChain-compatible wrapper with
  asymmetric compression (docs compressed, queries remain FP32)
- Added [rag] optional dependency group (langchain-core, numpy)

## GCP Scope Resolution (providers/gcp/)
- New GCPScopeResolver with project/folder/billing_account routing
- Folder scope: lists child projects via ResourceManager v3 API
- Billing account scope: lists linked projects via Cloud Billing API
- Resource label filtering across all GCP managers (compute, network, storage)
- Updated GCPAuthConfig with scope_type, scope_id, resource_labels fields

## Tests (31 new, 78 total)
- test_turbo_quant.py: cosine similarity >0.95, shape, dtype, seed
  reproducibility, error handling, LangChain wrapper delegation
- test_gcp_scope.py: mock-based API verification for all scope types,
  label filtering, error handling

## Website & PyPI
- Removed all emojis from index.html and landing/index.html
- Replaced with inline SVG icons and CSS gradient provider badges
- Updated pyproject.toml: version 0.1.4, homepage -> GitHub Pages,
  added google-cloud-resource-manager and google-cloud-billing deps
- Package builds and passes twine check