integrand: Integrand for Fourier transform
In gbasulto/mvdeconvolution: Multivariate kernel deconvolution

This is the integrand in the bivariate deconvolution formula, except for the exp(-itx) term.

1	integrand(t, EMethod, error, sigE, truncate, h, samp, columns, Kkernel)

`t`	A bivariate vector.
`EMethod`	Method used to approximate the characteristic function of the error. 1 or 2 if ecf of the char. fnc. of the kde is used, respectively.
`error`	Matrix with 2 columns, either with the pure sample of the error, or with the differences obtained from the panel data structure.
`sigE`	Bandwidth (2x2) matrix for the error. It is NULL by default and it is required only when EMethod = 2.
`truncate`	NULL by default. If the approximated characteristic function of the error is smaller than truncate, it will make the integrand zero. This provides numerical stability.
`h`	Bivariate vector with the bandwidth parameters.
`samp`	2-column matrix with the contaminated sample.
`columns`	Number of columns of the panel data. If there is one column, it assumes that instead panel data, a pure sample of the error is provided.
`Kkernel`	Kernel to be used. See 'ker' function.