v0.3.0

Changes:

• Added support for Ollama multi-modal models. Images can now be attached to a prompt or pasted in from the clipboard.
• No longer cap the max tokens on OpenAI.
• Stop Ollama from unloading models automatically.