GitLab

Add test for thor selections with fastest reco

Since the full reconstruction is functional now. It is not necessary anymore to return
a list of algorithms. If needed, the protoparticle creation can be scheduled explicitly
by inserting the datahandle into the control flow.
Explicit scheduling of the proto particle making has been removed in various places.

Gitlab CI authored Oct 05, 2020 and Nicole Skidmore committed Oct 05, 2020

patch generated by https://gitlab.cern.ch/lhcb/Moore/-/jobs/10267491

Sascha Stahl authored Jul 01, 2020 and Christopher Rob Jones committed Jul 01, 2020

patch generated by https://gitlab.cern.ch/lhcb/Moore/-/jobs/8863891

The functionality is implemented in the following files

* protoparticles.py includes functions to create charged and neutral
  protoparticles from reconstruction output.
* hlt2_global_reco.py sets up the whole reconstruction data flow. Its
  output can be used in particle making and selections.

Additionally, the HLT2 muon ID is added as it was missing before. It had
to be disabled again, as it is not thread-safe.

The global_reco file is needed because the ProtoParticle algorithms are
not functional, as they take input and mutate it. They must be then be
scheduled explicitly. The global_reco methods return the data handles of
the ProtoParticles containers as well as the algorithms that need to be
run, in order, to produce those outputs.

Because this is all rather brittle, there is a new test,
hlt2_protoparticles_baseline, which compares the data members of each
created ProtoParticle to a reference. The reference has been
cross-checked with a similar test from Brunel.

A second new test, hlt2_particles_baseline, checks that we can build
Particle objects from the ProtoParticles.