Migrate from Gemini to Gemma 3 (4B Parameters)

# Bounty Points: 25

## PR Instructions
Commits should be pushed to the **'local-model'** branch only. 📥

---

## **Status: Local LLM Integration Complete** 🚀

### **Summary of Change: Integrating Existing Solution to Main Codebase**

The architectural shift to support local inference via **Gemma 3 (4B parameters)** has been successfully implemented and tested in a separate branch/environment. This pull request integrates the completed solution into the main codebase.

This migration replaces the external **Gemini API** dependency, providing significant benefits:
* **Privacy:** Data processing is now **privacy-first** (data stays on the local machine).
* **Cost & Latency:** **Zero ongoing API cost** and significantly **lower latency** with full **offline capability**.
* **Autonomy:** Freedom from external API rate limits and dependencies.

The core backend flow remains unchanged; only the LLM interface layer has been modified to support a seamless switch between backends.

---