llama.cpp

model: add NVIDIA Nemotron-3-Puzzle-75B-A9B (NemotronHPuzzle) support

#25444

Merged