feat: `cortex run model(:gguf)`

[vansangpfiev]

Goals

• Cortex run should load and run model
• Cortex should run model with relevant parameters
• Cortex should display informative error messages if it fails

Tasklist

• cortex run model:llamacpp
• Ensure dylib issue is fixed bug: libengine.dylib not found #953
• Decision: should we uncouple engine install? (e.g. error message, directing user to install engine with separate command)
• Informative Error messages for major failure modes (e.g. OOM)
• Tests
• Docs

[vansangpfiev]

run is a chain command, it will do:

• models pull if model is not existed
• engines install if engine is not existed
• models start
• chat