feat: Support local LLM inference engines (Ollama, LM Studio / OpenAI-compatible API)

[CodeMaverick-143]

Feature Request

Add support for local AI models (e.g., Ollama running llama3, deepseek-coder, qwen, or LM Studio) so users can analyze private repositories locally without sending proprietary code to cloud endpoints.

Proposed CLI Options

# Using Ollama
insight . --provider ollama --model llama3

# Using custom OpenAI-compatible endpoint (LM Studio, vLLM)
insight . --provider local --api-base http://localhost:1234/v1 --model local-model

Proposed Architecture

Abstract the LLM provider in insight/detector.py:

• GeminiProvider: Uses Google Generative AI
• OllamaProvider: Connects to http://localhost:11434/api/generate
• OpenAICompatibleProvider: Connects to any OpenAI-compatible /v1/chat/completions endpoint