llama.cpp

cmake: use 1 more thread for non-ggml in CI

#8740

Merged