confused float and int in nvWavenetReference, setInputs

[nartes]

nv-wavenet/nv_wavenet_reference.cpp

Line 246 in 03f6957

memcpy(m_outputSelectors->data(), outputSelectors, m_maxSamples*m_maxBatch*sizeof(int));

Has no effect since input and output buffers are of float type (Matrix class).
Yet it is sizeof(int) for copying float data.