llama.cpp

fix(openvino): explicit ov::Tensor frees in ggml_backend_openvino_free

#21411

Closed