Fix BDT for the 'Hlt2CharmHadInclXicpToKmPpPipBDTTurbo' line

By looking at the 2018 validation sample, a bug was found in the look-up table of the line of 'Hlt2CharmHadInclXicpToKmPpPipBDTTurbo' line, in which the order of the last four variables was not set correctly. The background contribution from this trigger line from 2018 validation sample was shown to be very high and the incorrect look-up table is thought to be the problem.

I have already added the corrected look-up table into lhcb-datapkg/ParamFiles with the MR !13.

With the fixed look-up table, we re-test the rate, and the result are:

BDT threshold	0.15	0.20	0.25	0.3
Rate [Hz]	83.6 ± 13.6	52.0 ± 10.8	39.6 ± 9.3	15.4 ± 5.8

I then re-estimated the efficiency with different MC samples containing Xi_c+. At BDT>0.25, the efficiency is estimated to be:

Xi_c+ sources	Prompt	Xi_cc+	Xi_bc+	Xi_bc0
Eff [%]	3.438 ± 0.371	4.639 ± 0.755	5.334 ± 0.266	9.133 ± 0.452

The efficiency is improved by a factor of ~1.2 - ~2.5 (depending on the source of Xi_c+) compared to the corresponding cut-based line (Hlt2CharmHadXicpToPpKmPipTurbo), with the rate around 40Hz.

The CPU time is supposed to be the same, since variables and binnings are essentially the same as the previous look-up table (the bugfix only involves the change of order of the variables).