PWGHF: fix filling of Lc->pKpi background table filling in polarization analysis.

[mfaggin]

• change all bool columns to int8_t
• reorder template-argument declaration in template functions
• add column to keep track of real Lc->pKpi reconstructed signal
• enable table filling only w/o grouping by mcCollision, to avoid memory leaks in case of 3-prong candidates (not only background, but also signal) that have reassociated daughters, for which the particles do not belong to the current mcCollisions. For the signal, these are cases recovered by the track-to-collision-associator, but with mcCollision grouping in these cases a memory leak happens when retrieving the daughter-particle mother, because of the grouping of the particles table.