Fix calculations of one-strip cluster uncertainty

Fixing a bug in the calculations of the uncertainty of one-strip clusters. sTGC one-strip clusters aren't usually associated to good tracks. So the bug doesn't have large effect, but it should be fixed.

Edit: The changes in the STGC_Measurements container is expected, since the uncertainties of one-strip clusters were wrong due to the bug.

Tagging @pscholer, @dpizzi, @alaurier