Make FlightOps AI recruiter ready

[mitulpatel123]

What changed

• rebuilt the README around verifiable recruiter evidence, an end-to-end demo, API examples, architecture, tradeoffs, and roadmap
• added Mermaid diagrams to the README and architecture notes
• redirected the base URL to the interactive FastAPI documentation
• configured a writable ephemeral SQLite path and explicit FastAPI entrypoint for Vercel
• added an API test for the documentation redirect

Why

The original project worked but required a reviewer to infer its engineering value. This change makes the shipped backend, data, reliability, explainability, observability, and delivery evidence easy to evaluate without overstating unfinished ML or cloud work.

Impact

Recruiters and engineers can understand the project in approximately two minutes, run the full flow locally, and distinguish completed capabilities from planned production milestones. The project is also ready for a serverless demo deployment, with its ephemeral storage limitation documented explicitly.

Validation

• python -m pytest — 6 passed
• git diff --check
• GitHub Actions — passed
• production deployment — https://flightops-ai-mu.vercel.app/docs
• live checks — root 307 to /docs; health, docs, risk, summary, and metrics 200; ingestion 201