A theoretical question: Is it reasonable to use the same Theiler window as for diagonal structures, or even to use a Theiler window at all for the calculation of parameters like LAM or TT?

With the Theiler window, the diagonal structures around the LOI are ruled out of the analysis, but the effect on vertical structures is more complex: they are not ignored, but "chopped" or "split". So, if they are longer than the minimum length for counting them, the histogram of vertical structures is not just reduced, but shifted to lower values - and actually the total count of vertical structures might even increase!