a/ AliTRDCalibTask.cxx .h: one histo more to quantify the event selection if any problem.
b/ AliTRDCalibraVdriftLinear.cxx, .h: the number 1199 of entries has to be avoid with the robust fit from root TLinearFitter...otherwise it breaks...
c/ AliTRDPreprocessorOffline.cxx .h: I modify a bit the first implementation of Julian for the condition to mask or not (no entries OR gain above 15 rms away of mean gain over the chambers OR vdrift above 15 rms away of the mean vdrift over the chambers). I added some basic functions to produce automaticely in pass0 a chamber status. This can still change a bit...
d/ AliTRDCalDet.cxx .h: I introduce the robust rms used in c/. This removes the chambers very far away due to small anode voltages to have a more realistic rms.