WIP expose branch name to particle mapping.

The idea is to allow tuple tools to operate on more than one particle in the decay chain. In the example use case, I want to TISTOS a Kµ pair in a Bs→J/ψ(µµ)φ(KK) decay (such that using the tistos tuple tool on none of the particles Bs, J/ψ, φ, K+,K-,µ+,µ- would do the job).

The solution of buffering the branchname→const LHCb::Particle* relation certainly breaks any thread safety!

Please let me know what you think.