Add more tests of the new functor machinery
We should test the functor machinery more systematically.
As a first step, this could just mean defining (in Python) some lists of "functors that should work on a vertex", "functors that should work on a fitted track" and so on, and then configuring an algorithm to instantiate these functors for the various different "vertex", "fitted track", ... types that we use. This will check that the various template instantiations compile.
As a subsequent step, it would be good to check that the results of executing the functors also make sense. This would need some guaranteed-stable containers of tracks, vertices etc. that we could apply the functors to and compare the results to some reference.