llama.cpp

ggml : add graph tensor allocator

#2411

Merged