fix: also add tracks that failed smoothing to failed_tracks

[wdconinc]

Briefly, what does this PR introduce?

Since #1677 we remove tracks that failed extrapolation. We don't do that with tracks that failed smoothing. This PR also applies the same treatment to tracks that failed smoothing.

What kind of change does this PR introduce?

• Bug fix (issue #__)
• New feature (issue #__)
• Documentation update
• Other: __

Please check if this PR fulfills the following:

• Tests for the changes have been added
• Documentation has been added / updated
• Changes have been communicated to collaborators

Does this PR introduce breaking changes? What changes might users need to make to their code?

No.

Does this PR change default behavior?

No.

[github-actions]

Capybara summary for PR 1741

• rec_dis_10x100_minQ2=0_craterlake
• rec_dis_10x100_minQ2=1000_craterlake_tracking_only
• rec_dis_18x275_minQ2=0_craterlake_18x275
• rec_dis_18x275_minQ2=1000_craterlake_18x275
• rec_dis_5x41_minQ2=0_craterlake_5x41
• rec_e_1GeV_20GeV_craterlake
• rec_pi_1GeV_20GeV_craterlake
  ^{_{Last updated 2025-02-21T16:43-05:00 ddbe5ae}}

[veprbl]

I don't think we should be "hiding" tracks that failed smothing without a clear reason. If the tracks constituted by background, or have unreasonable parameters estimated, then we could reconsider. Would you agree?

[wdconinc]

I thought that ship had sailed when we decided to hide tracks that fail extrapolation? How is smoothing different?

[veprbl]

I thought that ship had sailed when we decided to hide tracks that fail extrapolation? How is smoothing different?

Tracks that fail extrapolation do not have a usable set of parameters, so we can't store them.

[wdconinc]

I thought that ship had sailed when we decided to hide tracks that fail extrapolation? How is smoothing different?

Tracks that fail extrapolation do not have a usable set of parameters, so we can't store them.

But we don't even attempt extrapolation when smoothing fails. If we want to allow tracks that failed smoothing, then we should still attempt extrapolation.

[veprbl]

I thought that ship had sailed when we decided to hide tracks that fail extrapolation? How is smoothing different?

Tracks that fail extrapolation do not have a usable set of parameters, so we can't store them.

But we don't even attempt extrapolation when smoothing fails. If we want to allow tracks that failed smoothing, then we should still attempt extrapolation.

Right, so in the current setup, smoothing has never ever failed, because if it would, that'd cause a segfault. If we apply your suggested change, if will be allowed to fail without interrupting anything.

[veprbl]

Should we pivot towards making this into a "fatal" exception?

[wdconinc]

Should we pivot towards making this into a "fatal" exception?

Define 'fatal'.

• It indicates something is fundamentally wrong with the geometry conversion to ACTS, so we should throw an exception and just fail the entire run.
• It is fatal for this track only, and we skip it entirely.

[veprbl]

Fatal as in this was never expected to happen and user is forced to report/confront this.