fix: Kalman Filter with ATLAS Stepper

[Corentin-Allaire]

Fix the issues related to the ATLASStepper so that it is usable with the KF :

• The pVector is the updated in update instead of resetState it seems this is the equivalent of the jacToGlobal in the EigenStepper
• The pVector is not updated when calling transportCovarianceToBound this would let to the covariance changing depending on how many time this method was called
• The ATLAS Stepper now create state with the transpose of its jacobian (which correspond to the same jacobian as the Eigen Stepper one)
• With this implementation the stepper curently doesn't work when encountering non-measurement surfaces, this is probably due to some kind of reinitialisation not being perform correctly ? This should be investigated.

And the PropagationCompareAtlasEigenConstant now also compares the covariances.

[codecov]

Codecov Report

Merging #1223 (ea587b7) into main (6c2807a) will decrease coverage by 0.25%.
The diff coverage is 58.07%.

@@            Coverage Diff             @@
##             main    #1223      +/-   ##
==========================================
- Coverage   47.89%   47.64%   -0.26%     
==========================================
  Files         375      375              
  Lines       19588    19624      +36     
  Branches     9214     9287      +73     
==========================================
- Hits         9382     9350      -32     
- Misses       3822     3885      +63     
- Partials     6384     6389       +5     

Impacted Files	Coverage Δ
Core/src/TrackFitting/GainMatrixSmoother.cpp	10.41% <0.00%> (-0.70%)	⬇️
Core/include/Acts/Propagator/AtlasStepper.hpp	62.05% <58.59%> (-6.46%)	⬇️

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

[asalzburger]

We may override the codecov, this is literally shifting code around, weird why this results in a negative codecov.

[Corentin-Allaire]

We may override the codecov, this is literally shifting code around, weird why this results in a negative codecov.

It seem to be related to some internal variable not beeing tested. But I am not fully sure and don't have much expertise with codecov...