Model switching and idle unload

Switching

hearthwork start --model X while the host is running switches models without stopping the host:

1. New requests wait in the queue.
2. Running requests finish (up to 2 minutes).
3. The prompt cache is saved.
4. llama-server restarts with the new model (auto context).
5. The queue resumes.

It refuses while requests are running unless you add --force, which cancels them after 10 seconds (their clients get a clear error). Only one switch runs at a time. hearthwork status shows switching to X (draining N) and loading X.

Idle unload

Idle unload is off by default.

hearthwork idle 30     # unload the model after 30 minutes without requests
hearthwork idle off    # turn it off

Unloading frees VRAM and RAM. The host keeps running, and the next request loads the model again. Clients just wait, and status says loading. The model is never unloaded while a request is running or waiting.

The work budget preset turns idle unload on (15 minutes).