Draft: Optimise GetFlavourTaggingParticles
GetFlavourTaggingParticles was doing redundant work of re-finding the PVs for B candidates for each flavour tagging particle. Per @graven's suggestion, this MR loops over the B candidates once and for each flaovur taggng particle, breaks the loop once it finds a PV that is better for it than the ones with B candidates.
This solves one half of the issue in #468.