Draft: Add LoKi adaptor ThOr functor

Towards Rec#169 (closed). Currently requires Rec!2429 (closed) but probably shouldn't; the Python code there should be moved here or the LoKi dependency moved to Rec.

To be addressed, off the top of my head:

• Tests. (I ran a local test and got identical results with F.LOKI("PT") versus F.PT, but should be formalised).
• Compatibility with the DV class of LoKi functors. These expect the owning algorithm to be derived from DVCommonBase so that they can call functions to e.g. get PVs and PV relations, but we expect to drop DVCommonBase support in the future. The algorithms we already have which operate on LHCb::Particle using ThOr functors do not derive from DVCommonBase.
• Adapters for combinations. The A series of LoKi functors, like AM, have a different signature. The adapter here is hard-coded for the signature of LHCb::Particle input (i.e. not combinations).
• Support 'preambulos', i.e. lists of strings of Python code to be executed before the functor code is evaluated.
  • TBD if we actually want this.