GitLab

* remove c code
* pack llama.cpp
* use request context for llama_cpp
* let llama_cpp decide the number of threads to use
* stop llama runner when app stops
* remove sample count and duration metrics
* use go generate to get libraries
* tmp dir for running llm

Co-Authored-By: Jeffrey Morgan <jmorganca@gmail.com>

Fixes #282

- default to embeddings enabled
- move embedding logic for loaded model to request
- allow embedding full directory
- close llm on reload

num_keep defines how many tokens to keep in the context when truncating
inputs. if left to its default value of -1, the server will calculate
num_keep to be the left of the system instructions

* resolves: https://github.com/jmorganca/ollama/issues/300 and
https://github.com/jmorganca/ollama/issues/282

* example usage:
```
ollama serve --port 9999 --allowed-origins "http://foo.example.com,http://192.0.0.1"
```

- read runner options from map to see what was specified explicitly and overwrite zero values