Add possibility to use Ollama server

[milkowski]

It would be great if Icebreacer could act as frontend UI for Ollama server and there was no need to download models separately by internal llama.cpp instance.

[hecrj]

We could add more options, but keep in mind that Ollama is actually a llama.cpp wrapper.

[milkowski]

Yes, but it is impossible to reuse its model storage in current form and that is the reason to have such option.