feat: GSF python example

[benjaminhuth]

This PR proposes a first version of an GSF example. The implementation of this required some bigger changes in the TrackFittingAlgorithm, since I wanted the TrackFittingAlgorithm to support different fitters with different options.

• I split the "algorithm-specific" options from the "fitter-specific" options. This now looks (in python) like this (for the Kalman Fitter):

  kalmanOptions = {
      "multipleScattering": True,
      "energyLoss": True,
      "reverseFilteringMomThreshold": reverseFilteringMomThreshold,
  }

  fitAlg = acts.examples.TrackFittingAlgorithm(
      level=acts.logging.INFO,
      inputMeasurements=digiAlg.config.outputMeasurements,
      inputSourceLinks=digiAlg.config.outputSourceLinks,
      inputProtoTracks=inputProtoTracks,
      inputInitialTrackParameters=smearAlg.config.outputTrackParameters,
      outputTrajectories="trajectories",
      directNavigation=directNavigation,
      pickTrack=-1,
      trackingGeometry=trackingGeometry,
      dFit=acts.examples.TrackFittingAlgorithm.makeKalmanFitterFunction(
          field, **kalmanOptions
      ),
      fit=acts.examples.TrackFittingAlgorithm.makeKalmanFitterFunction(
          trackingGeometry, field, **kalmanOptions
      ),
  )
  s.addAlgorithm(fitAlg)

• The extensions (with exception of the MeasurementCalibrator) live now inside the TrackFitterFunction, since there may be different extensions required by different fitters. The MeasurementCalibrator still must live in the TrackFittingAlgorithm, since it must be constructed with the measurements, which are obviously not known when constructing the TrackFitterFunction.

Besides this there are some minor changes in the GSF (moved some config structs from Acts::detail to Acts, ...), but should be uncontroversial

[paulgessinger]

Looks reasonable overall, I've added a few comments here and there.

At the python level, there could be some cleanup. I'm also wondering if at this point we should rename truth_tracking.py to make it obvious it runs the KF.

Aside from that: do you think you could look into running the GSF example from a pytest test, so we run it in the CI and do basic checks on the output? Do you have any kind of code / tooling to produce validation plots for the GSF? If so, that could potentially also be added and monitored by the CI.

[benjaminhuth]

Regarding the physics performance monitoring: In principle the fitter returns a Acts::KalmanFitterResult object, so maybe one could reuse some infrastructure there? I also have some custom plotting scripts, because I'm not so familiar with the ROOT things, but they should do similar things than the Root-Writer...
I will look into this in a follow-up PR if this is okay!

[benjaminhuth]

I did some renaming at this point as @paulgessinger mentioned above (truth_tracking.py -> kalman_tracks.py, gaussian_sum_filter.py-> gsf_tracks.py), and added the functions addKalmanTracks(...) as well as addGsfTracks(...).

I think this is now more in line with the other examples

[paulgessinger]

Changes look good to me in principle. There's some build failures right now though.

[codecov]

Codecov Report

Merging #1232 (f722fcd) into main (6e64a4c) will decrease coverage by 0.00%.
The diff coverage is 25.00%.

@@            Coverage Diff             @@
##             main    #1232      +/-   ##
==========================================
- Coverage   47.71%   47.71%   -0.01%     
==========================================
  Files         375      376       +1     
  Lines       19718    19722       +4     
  Branches     9264     9267       +3     
==========================================
+ Hits         9409     9410       +1     
- Misses       3920     3922       +2     
- Partials     6389     6390       +1     

Impacted Files	Coverage Δ
...re/include/Acts/TrackFitting/GaussianSumFitter.hpp	71.03% <ø> (ø)
Core/include/Acts/TrackFitting/detail/GsfActor.hpp	41.24% <ø> (ø)
Core/include/Acts/TrackFitting/GsfOptions.hpp	25.00% <25.00%> (ø)

📣 Codecov can now indicate which changes are the most critical in Pull Requests. Learn more

[paulgessinger]

Note that there are some CI issues that should be fixed by #1243.