Migrate trigger matching to using a tool to calculate DR

This is an underlying change required for a few more advanced things being discussed e.g. for egamma to match calo clusters rather than calibrated momenta or matching for LLPs. Both of these require being able to specialising the way the matching tool calculates how good a match is (for the default tool this is dR). A few questions I'd like some input on

• Should we migrate these interface classes to the TrigAnalysisInterfaces package
• The IMatchScoringTool interface is incredible generic - would it make sense to have this somewhere more general than the trigger domain?

I've also now added a scoring tool for egamma object - the way this ends up working does make me wonder if we actually should have one that runs a larger switch statement over object types - this could then run similar selections for muon objects (and then the switching in the TypedScoringTool is unnecessary.

Removing draft as this has been sitting here too long.