Add threadfriendly THistSvc

This is Charles' threadfriendly THistSvc that we've been using in our GaudiAna demonstrator. It can handle copies of histograms for cloned Algorithms in a multithreaded job and provides some means for merging them in the Algorithms finalize() method. There are also newly introduced shared Histograms, which are automatically locked internally with a LockedHandle.

I marked this as WIP, since the CI pipeline is failing in my fork (during apply-formatting-python) and I would be glad about some pointers on how to fix this properly. Any additional wishes of things I should do beyond that?

btw. this might be related to GAUDI-1258