cupy.cuda.runtime.mallocAsync# cupy.cuda.runtime.mallocAsync(size_t size, intptr_t stream) → intptr_t#