fix BPVDLS in thor

When implementing BPVDLS in Thor one line is wrongly commented out: int OK = W.Invert();. This results in a wrong BPVDLS calculation and too small trigger efficiency when testing BandQ detached dimuon lines.

After fixing this bug, the efficiency looks reasonable. A log file and option file is attached. Hlt2BandQ_oldJpsiToMuMuDetachedLine is the old LoKi version line and Hlt2BandQ_JpsiToMuMuDetachedLine is the ThOr-based on in current master branch. One can see that:

  LAZY_AND: Hlt2BandQ_JpsiToMuMuDetachedLineDecisionWithOutput                                                 #=889     Sum=462         Eff=|( 51.96850 +- 1.67565 )%|
   LAZY_AND: Hlt2BandQ_JpsiToMuMuDetachedLine                                                                  #=889     Sum=462         Eff=|( 51.96850 +- 1.67565 )%|

  LAZY_AND: Hlt2BandQ_oldJpsiToMuMuDetachedLineDecisionWithOutput                                              #=889     Sum=462         Eff=|( 51.96850 +- 1.67565 )%|
   LAZY_AND: Hlt2BandQ_oldJpsiToMuMuDetachedLine                                                               #=889     Sum=462         Eff=|( 51.96850 +- 1.67565 )%|

Hlt2DiMuonLines_detachedJpsi_new.py

log_Hlt2DiMuonLines_detachedJpsi_new

Based on the development of !2566 (merged) , necessary warning message is added when Invert does not work.