Can someone explain how the lower and upper slope bands are computed in scipy.stats.theilslopes?

Considering that theilslopes computes the slope as the median of all slopes between all possible paired values, if one defines alpha (confidence degree) equal to .95, does the computed upper bound value signify that 95% of the slopes between each pair of values are below this value?

I looked at the main reference

P.K. Sen, “Estimates of the regression coefficient based on Kendall’s tau”, J. Am. Stat. Assoc., Vol. 63, pp. 1379-1389, 1968.

but I didn’t find a clear description.