support sample.Mul(lhs, rhs).tri with 1D rhs

[joostvanzwieten]

Currently only sample.Mul(lhs, rhs).get_element_tri() is implement for 1D lhs. Unfortunately, when multiplying three 1D samples, X * Y * Z (or sampling a product of three 1D topologies), we get Mul(Mul(X, Y), Z), for which get_element_tri() is not implemented.