Optimize tensor loading #1129

Job	Run time
macOS-latest-cmake	2m 59s
ubuntu-latest-cmake	2m 9s
windows-latest-cmake (avx2, -DGGML_NATIVE=OFF -DGGML_AVX2=ON -DSD_BUILD_SHARED_LIBS=ON)	2m 21s
windows-latest-cmake (avx512, -DGGML_NATIVE=OFF -DGGML_AVX512=ON -DGGML_AVX=ON -DGGML_AVX2=ON -DS...	2m 14s
windows-latest-cmake (noavx, -DGGML_NATIVE=OFF -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF -DSD...	2m 11s
windows-latest-cmake (cuda12, -DSD_CUDA=ON -DSD_BUILD_SHARED_LIBS=ON -DCMAKE_CUDA_ARCHITECTURES=9...	35m 30s
windows-latest-cmake (vulkan, -DSD_VULKAN=ON -DSD_BUILD_SHARED_LIBS=ON)	5m 19s
windows-latest-cmake (avx, -DGGML_NATIVE=OFF -DGGML_AVX=ON -DGGML_AVX2=OFF -DSD_BUILD_SHARED_LIBS...	2m 43s
release	0s
	55m 26s

Provide feedback