[FEAT] Add healthcheck and resource limits for llm-service

[MikeNovikoff]

Feature description

Ensure the LLM service is robust in production by adding Docker healthchecks and sensible HTTP timeouts.

What is needed to be done

• docker-compose.yml: healthcheck block using Python HTTP client to hit /health
• quiz_core/internal/api/handlers/llm/generate.go: http.Client{Timeout: 120s}
• quiz_core/internal/api/handlers/llm/generate.go: http.MaxBytesReader on request body

Acceptance Criteria

• Docker healthcheck polls localhost:7000/health every 10s
• Go HTTP client timeout is set to 120 seconds
• Request body limited to 1 MB to prevent abuse
• Uploaded file extension restricted to .txt