Bug fix and addition of new training options for BDT readout

[anehrkor]

This PR fixes a bug where the decay mode 11 (3-prong+1Pi0) was not considered when reading the BDT output.

Moreover, new training options are added to the read out for upcoming trainings that include the Gottfried-Jackson angle as a variable, and no longer use the leading track chi2.

[roger-wolf]

Hi Alex,
this is for future MVA trainings? How much are current trainings touched by this? E.g. I would have expected a switch at cfg level that indicates whether to choose or not to choose the GJ-angle as a variable in the readout of the MVA. Or is it recognized automatically that for a given training this variable is not contained?

Also the bugfix: does this have an impact on the r/o of current MVA trainings?

Cheers,
Roger

[anehrkor]

Hi Roger,

current trainings are not touched by this at all since the user needs to decide which MVA option to use in his/her config file (see for example here).
Currently, no sqlite file with working points for the new variable has been uploaded to CondDB.

The bug fix has no impact on any trainings performed so far. It is however relevant for upcoming trainings with new MC, as the new tau decay mode is reconstructed from CMSSW 8_1_X onwards.

Cheers,
Alex