feat!: MultiTrajectory backends

[paulgessinger]

This PR decouples the interface of MultiTrajectory from the storage backend. The default storage backend is practically identical to the previous implementation, but it's now behind the interface. This change is intended to allow seamless integration with experiment specific EDM backends, like ATLAS' xAOD.

See these slides for some more information.

BREAKING CHANGE: The template parameters of CombinatorialKalmanFilter, KalmanFitter and GaussianSumFitter changes to

template <typename propagator_t, typename traj_t>
class CombinatorialKalmanFilter;
template <typename propagator_t, typename traj_t>
class KalmanFitter;
template <typename propagator_t, typename traj_t,
          typename bethe_heitler_approx_t = detail::BetheHeitlerApprox<6, 5>>
struct GaussianSumFitter;

Related classes (options, extensions, result) also change, gaining a traj_t template parameter. This parameter is the type of the memory backend.

[codecov]

Codecov Report

Merging #1262 (96894f6) into main (e72db5d) will increase coverage by 0.36%.
The diff coverage is 58.18%.

@@            Coverage Diff             @@
##             main    #1262      +/-   ##
==========================================
+ Coverage   47.46%   47.83%   +0.36%     
==========================================
  Files         375      380       +5     
  Lines       19842    20149     +307     
  Branches     9293     9371      +78     
==========================================
+ Hits         9418     9638     +220     
- Misses       4045     4063      +18     
- Partials     6379     6448      +69     

Impacted Files	Coverage Δ
.../include/Acts/EventData/MultiTrajectoryHelpers.hpp	50.00% <ø> (ø)
Core/include/Acts/EventData/TrackStatePropMask.hpp	100.00% <ø> (ø)
...re/include/Acts/TrackFitting/GainMatrixUpdater.hpp	10.00% <0.00%> (ø)
...e/Acts/TrackFitting/detail/KalmanUpdateHelpers.hpp	15.25% <0.00%> (-6.78%)	⬇️
Core/src/EventData/TrackStatePropMask.cpp	0.00% <ø> (ø)
Core/src/TrackFitting/GainMatrixSmoother.cpp	7.69% <7.69%> (-2.95%)	⬇️
Core/src/TrackFitting/GainMatrixUpdater.cpp	20.00% <16.66%> (+8.88%)	⬆️
Core/include/Acts/TrackFitting/detail/GsfUtils.hpp	39.24% <24.32%> (-13.15%)	⬇️
Core/src/TrackFinding/MeasurementSelector.cpp	33.33% <33.33%> (+0.90%)	⬆️
.../include/Acts/TrackFinding/MeasurementSelector.hpp	38.09% <36.06%> (-61.91%)	⬇️
... and 27 more

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

[paulgessinger]

Following @timadye's suggestions, I'll try to replace void* with std::any to get at least type checking.

[tboldagh]

Just a few comments (more questions realy).
One thing that I missed to spot, and was curious was creation of new track states
in CKF, KF or GSF and how this changes with backend?

[paulgessinger]

FYI I'm working on a follow-up update that uses CRTP instead of virtual functions.

[andiwand]

just took a quick look - will follow up later

[andiwand]

sorry, I could not provide valuable feedback but highlighted some todo / fixme / commented code

[paulgessinger]

Thanks @andiwand, mostly things I missed when cleaning up. Should all be addressed with the most recent commit.

[paulgessinger]

I updated this with a merge commit, but the approval was dismissed. @tboldagh or @andiwand can you reapprove?