Bugfix: SE transformer + link with validation (#1567)

[mgovers]

Fixes #1567

This supercedes #1571 and #1572 .

Rationale

The original repro case was missing 2 things:

• Running those exact validation cases on version 1.13.146 (the final working version before introduction of the bug in [BUG] *WLS Asymmetrical State Estimation diverges and applies inconsistent Phase/Line scaling (Regression since >= 1.13.147)* #1567 ) resulted in a failure on certain platforms, therefore impeding a stable search for the bug.
• The test case did not contain a node injection sensor.
  • While that is not an issue in the current main, as there is only one relevant code path, Merge links into nodes: link registration refactoring (no functional change) #1550 will have different behavior, depending on whether the input data contains node injection sensors or not
  • Node injection sensors are deprecated and will be removed in v2.
  • If the input data contains no node injection sensors, links are separately handled and [BUG] *WLS Asymmetrical State Estimation diverges and applies inconsistent Phase/Line scaling (Regression since >= 1.13.147)* #1567 is automatically resolved. This will be the only remaining behavior in v2.
  • If there are node injection sensors, the original behavior remains, so the bug also needs to be resolved there.
  • As a result, Merge links into nodes: link registration refactoring (no functional change) #1550 will require repro cases for both with and without node injection sensors.

See also 123e0e5#r3948457387

Verification of validity

• Create validation cases
  • without node injection sensor
    • ILSE
    • NRSE
  • with node injection sensor
    • ILSE
    • NRSE
• Verify that all cases pass on v1.13.146 on all platforms + architectures
  • branch pgm/experiment/issue-1567 based off v1.13.146 contains exactly these validation cases
  • Manual CI run in https://github.com/PowerGridModel/power-grid-model/actions/runs/34210873087
• Verify that the validation cases reproduce [BUG] *WLS Asymmetrical State Estimation diverges and applies inconsistent Phase/Line scaling (Regression since >= 1.13.147)* #1567 on main
  • This PR based off main contains exactly the same validation cases
  • All cases are marked xfail, raising a SparseMatrixError
  • Verify that all tests pass (regular CI)

[Jerry-Jinfeng-Guo]

One question remaining (not related to the merging of this PR) is why the original incomplete repro case passes on the link registration brach

[figueroa1395]

One question remaining (not related to the merging of this PR) is why the original incomplete repro case passes on the link registration brach

@Jerry-Jinfeng-Guo this is because in #1550 and in the original test, we didn't have a node injection sensor, so the new code path was being triggered. In the new code path the topology is reduced, where links are being "hidden". The reason the reduced topology didn't trigger this is because the issue appears only when links and branches are stored together and the wrong branch/edge dispatch fails there, as it's expecting something different.

[sonarqubecloud]

Quality Gate passed

Issues
5 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud