chore: ⬆️ Update ggml-org/llama.cpp to `4f37f519722aa3242eecb7649466b4a4a2d6d6da` #15288

Job	Run time
image-build (hipblas, linux/amd64, false, -hipblas, rocm/dev-ubuntu-24.04:7.2.1, ubuntu-latest, -... / reusable_image-build	10m 3s
image-build (cublas, 12, 8, linux/amd64, false, -gpu-nvidia-cuda-12, ubuntu-latest, ubuntu:24.04,... / reusable_image-build	9m 28s
image-build (vulkan, linux/amd64, amd64, false, -vulkan-core, ubuntu-latest, ubuntu:24.04, --jobs... / reusable_image-build	32m 46s
image-build (cublas, 13, 0, linux/amd64, false, -gpu-nvidia-cuda-13, ubuntu-latest, ubuntu:22.04,... / reusable_image-build	8m 33s
image-build (sycl, linux/amd64, false, intel/oneapi-basekit:2025.3.2-0-devel-ubuntu24.04, sycl, u... / reusable_image-build	6m 55s
image-build (cublas, 13, 0, linux/arm64, false, -nvidia-l4t-arm64-cuda-13, ubuntu:24.04, ubuntu-2... / reusable_image-build	6m 7s
image-build (vulkan, linux/arm64, arm64, false, -vulkan-core, ubuntu-24.04-arm, ubuntu:24.04, --j... / reusable_image-build	6m 34s
	1h 20m 26s

Provide feedback