Ollama Support

[mik-tf]

Summary

• Adds comprehensive Ollama integration for running AI models locally, providing users with a privacy-focused, zero-cost alternative to cloud APIs.
• Closes Add Local AI Support #11

Key Features

• Multi-Provider Architecture: Switching between Google Gemini (cloud) and Ollama (local)
• Complete Ollama Integration: Automatic service detection, model management, and tool calling support
• Enhanced UI: Provider selection with dynamic model lists and contextual settings
• Robust Implementation: OpenAI-compatible API with error handling and conversation persistence
• Comprehensive Documentation: Step-by-step setup guides for native (Ollama) and Docker installations

Files Changed

• app.go - Provider switching and Ollama initialization
• ollama.go - Ollama provider implementation with tool calling
• Settings.svelte - UI for provider/model selection
• Docs.svelte - Complete Ollama setup documentation

Benefits

• Zero API Costs: Unlimited local AI usage
• Complete Privacy: Conversations never leave the machine
• High Performance: Instant responses with local processing (If good machine...)
• Easy Setup: Simple installation with comprehensive guides

Tests

• Base tests as I don't have a strong AI-ready machine locally

[mik-tf]

Converting to draft and will make another PR without conflict.