Add the EfficiencyFilter algorithm and tests running it

[jmcarcell]

BEGINRELEASENOTES

• Add an EfficiencyFilter, an algorithm that will take an EDM4hep collection and return a subset collection with a certain efficiency. The Exact parameter can be used to control whether the efficiency is exactly that number of simply
  each element is kept with Efficiency probability.
• Add a test that runs the EfficiencyFilter both in the exact version (to check that exactly Efficiency * length elements are saved) and in the non-exact version

ENDRELEASENOTES

It's also an example of a transformer that works on a single arbitrary collection, of which we didn't have any examples before.

See key4hep/k4Reco#27 (comment)

[jmcarcell]

@mahmoudali2 can you try to use this for simulating the efficiency? You would need to build k4FWCore with this pull request and then you can run DDPlanarDigi without building since you wouldn't need key4hep/k4Reco#27. This has the advantage that you can know exactly which hits are lost in DDPlanarDigi since you know which hits remain after the efficiency cut.

[m-fila]

This won't work with the data-model extensions, right?

[jmcarcell]

This won't work with the data-model extensions, right?

No, like any algorithm we create here, it can't know about the extensions. It's always the case with any algorithm, no?

[tmadlener]

I am wondering whether we can have a somewhat lightweight pseudo plugin mechanism here that makes it possible to add another type list downstream where extensions are defined. OTOH, not having all the functionality for all possible extensions probably also serves as a good incentive for people to actually go use EDM4hep.

[BrieucF]

I am wondering whether we can have a somewhat lightweight pseudo plugin mechanism here that makes it possible to add another type list downstream where extensions are defined. OTOH, not having all the functionality for all possible extensions probably also serves as a good incentive for people to actually go use EDM4hep.

I would argue here that since we have the policy to integrate in EDM4hep new data types only if they have been thoroughly tested, we'd rather provide the needed functionalities to fully integrate extensions into realistic workflows.