[FEAT]: Local model loading status

What would you like to see?

For local providers like Ollama, LMStudio, Lemonade there is a way to trivially detect if the model we are about to start running inference on is currently loaded.

In a fresh agent session, when we know the model is not loaded, we should report a "Loading model into memory" status that reports in the chain-of-activity UI.

On some systems this loading can take a long time and so to at least alert the user of this delay or loading state we should report this item. If the model is loaded then don't push anything.