Optionally refit ITS outward seeding with inward refit result

[shahor02]

No description provided.

[github-actions]

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2023-pbpb-apass4
async-2023-pp-apass4
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0
async-2024-PbPb-apass1
async-2024-ppRef-apass1
async-2024-PbPb-apass2
async-2023-PbPb-apass5

[alibuild]

Error while checking build/O2/fullCI_slc9 for 91a92fc at 2026-01-16 00:19:

## sw/BUILD/O2-latest/log
/sw/SOURCES/O2/14962-slc9_x86-64/0/Detectors/ITSMFT/ITS/tracking/GPU/cuda/TrackingKernels.cu(366): error: identifier "saveInw" is undefined
ninja: build stopped: subcommand failed.

Full log here.

[f3sch]

@shahor02, thanks! Did it bring any improvements if the default is set to true?

[shahor02]

@shahor02, thanks! Did it bring any improvements if the default is set to true?

I still have to run with both options, then to compare the outputs. Will change the default if needed.

[alibuild]

Error while checking build/O2/fullCI_slc9 for a4ed635 at 2026-01-16 10:16:

## sw/BUILD/O2-latest/log
/sw/SOURCES/O2/14962-slc9_x86-64/0/Detectors/ITSMFT/ITS/tracking/GPU/cuda/TrackingKernels.cu(343): error: identifier "savInw" is undefined
ninja: build stopped: subcommand failed.

Full log here.

[alibuild]

Error while checking build/O2/fullCI_slc9 for b9bf5a9 at 2026-01-17 20:01:

## sw/BUILD/O2-latest/log
/sw/SOURCES/O2/14962-slc9_x86-64/0/Detectors/ITSMFT/ITS/tracking/GPU/cuda/TrackingKernels.cu(366): error: identifier "saveInw" is undefined
ninja: build stopped: subcommand failed.

Full log here.

[shahor02]

Disable refit by default, as it does not show apparent advantage (red solid: with refit required, blue diamonds: w/o refit).
At the same time, in both cases, I see some degradation wrt improvement from the better linearization originally achieved in December (magenta circles). The black solid is for the old data w/o proper linearization.
At the moment, I have not found the reason for the degradation...

[f3sch]

Thanks, indeed weird why the blue is also degrading, should not be related to the PR though... Besides the recent changes in the clusterizer which should be good, nothing much changed besides this e12ca36 (although this should be valid....)