llama.cpp

ggml : build backends as libraries

#10256

Merged