Normalize API contract, fix scoring keys, and remove tracked bytecode

[MohammedAlkindi]

Motivation

• Previous large patch left duplicated/contradictory fragments between schema aliases, scoring weight names, and frontend consumption fields; the change aligns them into a single coherent contract.
• Tracked Python bytecode and virtualenv files cluttered the repo and risked re-introducing inconsistent behavior; they should be ignored and untracked.

Description

• Unified Pydantic model configuration to v2 style using model_config = ConfigDict(populate_by_name=True) for request/response models so alias handling is explicit and consistent across the API (backend/app/models/schemas.py).
• Normalized scoring weight keys in the engine and all references so BASE_WEIGHTS keys match the score breakdown component names (favorite_overlap, avoid_terms) and updated scoring logic to use the new keys (backend/app/scoring/engine.py).
• Simplified frontend summary rendering to a single deterministic path and removed transitional fragments so the UI text uses meta.summary and meta.appliedRefinements consistently (frontend/main.js).
• Added a valid example JSON response to the README.md that matches the backend/frontend contract (camelCase response aliases).
• Updated .gitignore to include .venv and backend/.venv and removed tracked __pycache__/ and .pyc artifacts from the repository.

Testing

• Ran python -m compileall backend/app to ensure backend code compiles; result: success.
• Ran node --check frontend/main.js to validate frontend JS syntax; result: success.
• Performed a lightweight contract smoke check by creating a RecommendationResult via to_result(...).model_dump(by_alias=True) and confirmed the serialized keys include imageUrl, detailsUrl, scoreBreakdown, and matchReasons; result: success.
• Verified tracked bytecode enumeration and removed files, then committed changes; removal operation and commit succeeded.

---

Codex Task