De-trending variables algorithm

This function uses as input a matrix whose column are different time series. Than it controls for each one if they are stationary or not, using an augmented Dickey-Fuller test for unit root. if the test fails and some variables follow a stationary trend, then the algorithm works on differences until the variable became stationary. Moreover there is a vector "counter", which says how many time the difference has been performed.
Differerencing is obtained by taking log-differences