feat: add Ollama local LLM support alongside Gemini

[ashishpatel26]

Introduces a pluggable LLM provider system so users can run the app fully offline using a local Ollama server instead of Google Gemini.

• Add src/services/gemini.service.js (extracted Gemini implementation)
• Add src/services/ollama.service.js (Ollama /api/chat client with vision support for screenshot analysis and conversation history)
• Convert src/services/llm.service.js into a dynamic provider router that delegates to gemini or ollama based on LLM_PROVIDER env var / runtime config — no restart required when switching
• Extend src/core/config.js with llm.provider and llm.ollama config block
• Add LLM_PROVIDER, OLLAMA_BASE_URL, OLLAMA_MODEL, OLLAMA_VISION_MODEL to env.example
• Add "AI Provider" selector and Ollama settings section to settings.html; Gemini/Ollama panels toggle based on selection
• Wire up new IPC handlers in main.js: get/set-llm-provider, get-ollama-status, test-ollama-connection
• Expose new IPC methods in preload.js