docs: add demos, benchmarks, and four new user guide pages

[nickprock]

Summary

• Added a Demo Notebooks section to README.md (after Quick Start) listing all 24 notebooks in demo/ grouped by category: Query Languages, Embedders, Vector Backends, and Features & Patterns.
• Created docs/demos.md — a dedicated index page for the demo notebooks with GitHub links, grouped by the same categories. Added as a top-level nav entry under Getting Started.
• Created docs/benchmarks.md — an index of all scripts in experiments/ grouped by topic (Latency, Throughput & Ingestion, Quality & Tuning, Cost & Observability, Dataset Generation). Added as a top-level nav entry under Getting Started.
• Created docs/user_guide/framework_integrations.md — full documentation for integrating Medha with LangChain (BaseCache), LlamaIndex (Workflow @step), and Haystack (@component), derived from demo/13_framework_integrations.ipynb.
• Created docs/user_guide/multi_tenancy.md — covers six multi-tenancy patterns (per-application isolation, per-customer isolation, shared embedder, concurrent fan-out, per-tenant stats, per-tenant backend selection), derived from demo/09_multi_tenant.ipynb.
• Created docs/user_guide/parameter_extraction.md — documents the ParameterExtractor cascade (Regex → GLiNER → spaCy → Heuristics), with usage examples and a backend selection guide, derived from demo/05_ner_spacy_vs_gliner.ipynb.
• Created docs/user_guide/security.md — documents all security settings (max_question_length, max_file_size_mb, allowed_file_dir, qdrant_api_key as SecretStr, PostgreSQL identifier validation) with a recommended production configuration.
• Updated mkdocs.yml nav accordingly: Getting Started is now a section with Overview / Demos / Benchmarks; four new pages added to the User Guide.

Notes

No Python source changes. No new dependencies. All new pages link back to the corresponding notebook or script in the repository.