Exploring Spiking Neural Networks for Binary Classification in Multivariate Time Series at the Edge