Improvement in GUI settings for OpenAI compatible (Ollama) LLM API

[osering]

Proposal adding "api_url" and "model (name)" fields under "Ollama" subtitle under Assistant Configuration. In this case it could let in gui configure other than 127.0.0.1 host (like usually needed 0.0.0.0) and other than 11434 port, therefore letting use not only Ollama, but also TextGen WebUI and other local backends, which use different ports and paths, but more or less common OpenAI compatible API.