Avoid duplication of statistical uncertainties (for smoothed b-tagging calibration objects), see AFT-483

This update fixes an issue with smoothed b-tagging calibration scale factor objects. In these cases, the statistical uncertainties are stored as variations, just like systematic uncertainties. However, they are also still added to the main histogram, for easy evaluation (presumably limited to plotting). This leads to a duplication, at least in the SFEigen uncertainty model.

I've fixed this by flagging these cases, by looking for variations having a name starting with "Stat_Nuis", and in such cases zeroing the statistical uncertainties covariance matrix. Conceptually this is not ideal as this leads to statistical and systematic uncertainty covariance matrices that individually are incorrect; however, the user only ever gets to see their sum, which is correct.

In addition, I have made two other changes:

1. addition of a "VerboseCDITool" Property to the BTaggingEfficiencyTool (set to true by default), which can be used to disable any regular printout by the CalibrationDataInterfaceTool code
2. update of the BTaggingEfficiencyToolTester code to use the latest CDI file and dated jet collection