refactor: Add nonlinear correction in KF

[XiaocongAi]

This PR refactors a few methods of the stepper related with transforming from free to bound parameters. It adds the option for correction of non-linear effects during the transformation using a few sigma points (https://doi.org/10.1117/12.280797).

[codecov]

Codecov Report

Merging #1233 (595c3bf) into main (0b8796e) will decrease coverage by 0.08%.
The diff coverage is 37.69%.

@@            Coverage Diff             @@
##             main    #1233      +/-   ##
==========================================
- Coverage   47.98%   47.89%   -0.09%     
==========================================
  Files         373      375       +2     
  Lines       19484    19588     +104     
  Branches     9148     9214      +66     
==========================================
+ Hits         9349     9382      +33     
- Misses       3804     3822      +18     
- Partials     6331     6384      +53     

Impacted Files	Coverage Δ
Core/include/Acts/Propagator/EigenStepper.hpp	68.42% <ø> (ø)
...re/include/Acts/Propagator/StraightLineStepper.hpp	67.85% <ø> (ø)
Core/src/Propagator/detail/CovarianceEngine.cpp	47.44% <15.00%> (-4.63%)	⬇️
...c/EventData/CorrectedTransformationFreeToBound.cpp	32.46% <32.46%> (ø)
Core/include/Acts/TrackFitting/KalmanFitter.hpp	45.07% <44.44%> (+0.41%)	⬆️
...e/Acts/TrackFitting/detail/KalmanUpdateHelpers.hpp	22.03% <50.00%> (ø)
.../include/Acts/Propagator/MultiEigenStepperLoop.hpp	70.37% <60.00%> (+0.18%)	⬆️
Core/src/Propagator/StraightLineStepper.cpp	68.42% <60.00%> (-1.03%)	⬇️
Core/include/Acts/Propagator/AtlasStepper.hpp	68.51% <66.66%> (ø)
...Data/detail/CorrectedTransformationFreeToBound.hpp	100.00% <100.00%> (ø)
... and 5 more

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

[XiaocongAi]

Hi @asalzburger , it's the PR for the non-linear KF. Could you help have a look?

[asalzburger]

Very nicely integrated, please check if the introduction of a small struct that carries a bool plus the unable parameters wouldn't make it even a step clearer - I suppose so!

[XiaocongAi]

Hi @asalzburger , thank you a lot for the suggestions. I should have resolved all the comments.

[asalzburger]

A few more things: Paul's suggestion of not auto-constructing from pool & the moving out of the detail namespace.

[XiaocongAi]

Hi @asalzburger @paulgessinger , I have resolved your comments. Could you have another look?

[asalzburger]

Comments are addressed, there is still a failure in the CI to check.

[asalzburger]

Hi @XiaocongAi - there are build failures, due to a missing test update & a mismatch in the concept, please resolve these ones and I will reapprove.

[paulgessinger]

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

[paulgessinger]

I would strongly prefer we make the FreeToBoundCorrection constructor from bool explicit like

explicit FreeToBoundCorrection(bool apply_);

See comment in the changeset.

[asalzburger]

From my side this is ok, let's see if the CI runs through and @paulgessinger is happy as well.

[XiaocongAi]

Hi @paulgessinger , thank you for your further comments. I have made the change. Do you think you can approve and get it in?

[paulgessinger]

LGTM as far as I can tell at this point!