RICH Future : Fix build platform differences.

This MR improves the RICH MC utility classes to better support cases where a single Track can have multiple MCParticles associated to.

Previously I would simply pick the best, based on the weight. When the weights where the same, the first association found was used. It turns out this caused differences between the opt and dbg builds due to the fact the Relations framework does not garantee the order of elements in the relations tables will be the same across builds (or even different runs for that matter) due to the fact sorting on memory location is performed. I am discussing this with @ibelyaev as I suspect this can (and should) be improved.

There are also a few other little (essentially cosmetic) changes to other RICH utilities.