TRT: fix an issue that allowed >=100% efficiencies to propagate into post-processing-dependent TRTViewer color maps
requested to merge ysmirnov/athena:dev_Yury_TRTViewerColorMapsJune2024_valuesOfGreaterThanOrEqualToOneHundredPercentFedIntoColorMaps into 24.0
When the outputs of parallel online-DQM jobs (after post-processing) are merged together, so are our TH1Fs with efficiencies along the Y axis. The efficiencies are smaller than 1 in a particular histogram in each of the output ROOT files before the merging, but after hadding these files this results in efficiency values >> 1. Switching the type of the object returned by the post-processing from TH1F to TProfile is supposed to fix the issue. The reason for replacing the SetBinContent() calls with Fill() is explained here. Tagging @cgrefe.