subprocess llama.cpp server (#401)

* 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

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "llm/llama.cpp/ggml"]
+	path = llm/llama.cpp/ggml
+	url = https://github.com/ggerganov/llama.cpp.git