Follow up from !1467

I can still reproduce the overlap reports from !1467 (merged) on my old stack but on a newer stack I see (after adding back the hlt2_pp_default cache):

Moore_FunctorCache_Hlt2_options_hlt2_pp_default.components has 1445 functors, let's check overlap:
Moore_FunctorCache_Hlt2_options_hlt2_pp_default.components has 700 functors overlap with Moore_FunctorCache_Hlt2_options_hlt2_pp_thor.components
Moore_FunctorCache_Hlt2_options_hlt2_pp_thor.components has 1597 functors, let's check overlap:
Moore_FunctorCache_Hlt2_options_hlt2_pp_thor.components has 700 functors overlap with Moore_FunctorCache_Hlt2_options_hlt2_pp_default.components
Moore_FunctorCache_Hlt2_options_hlt2_pp_thor.components has 1 functors overlap with Moore_FunctorCache_Hlt1_hlt1_pp_default.components
Moore_FunctorCache_Hlt1_hlt1_pp_default.components has 23 functors, let's check overlap:
Moore_FunctorCache_Hlt1_hlt1_pp_default.components has 1 functors overlap with Moore_FunctorCache_Hlt2_options_hlt2_pp_thor.components

First hint is that pp_thor is getting vertices from /Event/TrackBeamLineVertexFinderSoA/OutputVertices while hlt2_all_lines or the old pp_default have /Event/RecV1ToPVConverter/OutputVertices as location in their functors.

Not sure what exactly is going on, but maybe the reconstruction diverged between the two?

I'm off for a week so logging this in case someone has time to look into this, otherwise I'll do it when I'm back.

/cc @lhcb-rta-maintainers this is causing hlt2_all_lines and similar tests to fallback onto JIT compilation for a rather large amount which causes timeouts when the test is run on a slow machine. If it becomes a bigger problem, we need to add back one of those options to the MooreCache.