llama.cpp

ggml : move common CPU backend impl to new header

#9509

Merged