Replace with cleaned vector-only implementation

v2.0: Full float32 precision pipeline, headDim=64 support, 448/448 production tests passed

Delete tests/verify.py with huggingface_hub

Delete tests/test_pytorch_compare.py with huggingface_hub

Delete tests/test_cube.cpp with huggingface_hub

Delete tests/test_correctness.py with huggingface_hub

Delete tests/test_aclnn.cpp with huggingface_hub

Delete tests/gen_data.py with huggingface_hub

Delete tests/benchmark_msprof.py with huggingface_hub

Delete tests/benchmark.py with huggingface_hub

Delete tests/CMakeLists_vector.txt with huggingface_hub

Delete tests/CMakeLists_cube.txt with huggingface_hub

Delete smallq_flash_attention_cube/op_kernel/smallq_flash_attention_cube_impl.h with huggingface_hub

Delete smallq_flash_attention_cube/op_kernel/smallq_flash_attention_cube.cpp with huggingface_hub

Delete smallq_flash_attention_cube/op_host/smallq_flash_attention_cube_tiling.h with huggingface_hub

Delete smallq_flash_attention_cube/op_host/smallq_flash_attention_cube_tiling.cpp with huggingface_hub

Delete smallq_flash_attention_cube/op_host/smallq_flash_attention_cube_proto.cpp with huggingface_hub

Delete smallq_flash_attention_cube/op_host/smallq_flash_attention_cube_def.cpp with huggingface_hub

Delete smallq_flash_attention_cube/README.md with huggingface_hub

Delete smallq_flash_attention_cube/CMakeLists.txt with huggingface_hub

Delete smallq_flash_attention/op_kernel/smallq_flash_attention_impl.h with huggingface_hub

Delete smallq_flash_attention/op_kernel/smallq_flash_attention.cpp with huggingface_hub

Delete smallq_flash_attention/op_host/smallq_flash_attention_tiling.h with huggingface_hub

Delete smallq_flash_attention/op_host/smallq_flash_attention_tiling.cpp with huggingface_hub

Delete smallq_flash_attention/op_host/smallq_flash_attention_proto.cpp with huggingface_hub

Delete smallq_flash_attention/op_host/smallq_flash_attention_def.cpp with huggingface_hub

Delete smallq_flash_attention/README.md with huggingface_hub

Delete smallq_flash_attention/CMakeLists.txt with huggingface_hub

Delete build.sh with huggingface_hub

Delete README.md with huggingface_hub

Delete CMakeLists.txt with huggingface_hub

Upload tests/test_pytorch_compare.py with huggingface_hub

Upload folder using huggingface_hub

initial commit