GitLab

* app: gracefully shut down `ollama serve` on windows

* fix linter errors

* bring back `HideWindow`

* remove creation flags

* restore `windows.CREATE_NEW_PROCESS_GROUP`

This should resolve a number of memory leak and stability defects by allowing
us to isolate llama.cpp in a separate process and shutdown when idle, and
gracefully restart if it has problems.  This also serves as a first step to be
able to run multiple copies to support multiple models concurrently.

This should unmask some failure modes that likely
show up in app logs as unmarshal errors

default client already handles proxy

Also fixes a few fit-and-finish items for better developer experience

This focuses on Windows first, but coudl be used for Mac
and possibly linux in the future.