Import IdentityOperator from SciMLOperators in Core tests

[ChrisRackauckas-Claude]

Ignore until reviewed by @ChrisRackauckas.

Summary

• Import IdentityOperator directly from its owning package, SciMLOperators, in test/Core/traits.jl.
• Exercise the existing _isidentity_struct regression assertion with that owner import instead of reaching through SciMLBase.IdentityOperator.
• Keep the change test-only; LinearSolve already directly depends on SciMLOperators.

Root cause

SciMLBase commit b9c3507a2670fae804bf627ca32354efeee8dc15 (#1472) removed the blanket SciMLOperators re-export. Its parent, 7b806c50a9e0419d525374d8b716aa51918876b8, still exposes SciMLBase.IdentityOperator; the first-bad commit does not. SciMLBase #1476 explicitly keeps the SciMLOperators names dropped because SciMLOperators owns and documents them.

SciMLOperators v1.25.1 publicly exports IdentityOperator, provides its docstring, and renders it in docs/src/premade_operators.md, so the test can use the owner API directly.

Verification

• The unchanged current-main GROUP=Core harness is currently masked earlier by the open Import FunctionOperator from SciMLOperators in Core tests #1127 FunctionOperator regression.
• On a detached current-main + Import FunctionOperator from SciMLOperators in Core tests #1127 validation stack with SciMLBase v3.40.1, the official Core harness passed Default Alg Tests 109/109 and ForwardDiff Overloads 128/128, then reproduced Traits as 5 passes and one UndefVarError at SciMLBase.IdentityOperator.
• Focused fixed test: Traits 6/6.
• On the same validation-only stack with this patch, the complete official GROUP=Core harness passed through SpecializingFactorizations.
• Whole-repository Runic check exited 0.
• git diff --check exited 0.

#1127 is present only in the detached validation stack so the harness can reach Traits; this PR contains only the IdentityOperator owner-import correction.

[ChrisRackauckas-Claude]

Investigation/validation scratchpad:

• LinearSolve baseline: ed7bcb014f03d15f50e9a5cf86f8a314ab2d248d.

• The unchanged baseline's official GROUP=Core run with SciMLBase v3.40.1 stops first in test/Core/default_algs.jl at the open Import FunctionOperator from SciMLOperators in Core tests #1127 FunctionOperator regression. This is why the IdentityOperator reproduction uses Import FunctionOperator from SciMLOperators in Core tests #1127 as a validation-only prerequisite.

• Formal SciMLBase boundary:

  • 7b806c50a9e0419d525374d8b716aa51918876b8: SciMLBase.IdentityOperator is visible.
  • b9c3507a2670fae804bf627ca32354efeee8dc15 (#1472): first bad commit; the binding is not visible.

• SciMLOperators v1.25.1 owner audit:
  (owner_public = true, owner_exported = true, owner_docstring = true, rendered_docs = true).

• Current-main + Import FunctionOperator from SciMLOperators in Core tests #1127 official Core reproduction:

  • Default Alg Tests: 109/109.
  • ForwardDiff Overloads: 128/128.
  • Traits: 5 passes, 1 error, UndefVarError: IdentityOperator not defined in SciMLBase.

• Focused fixed result:

  Test Summary: | Pass  Total  Time
  Traits        |    6      6  0.1s
  

• Current-main + Import FunctionOperator from SciMLOperators in Core tests #1127 + this patch, official GROUP=Core result:

  • Default Alg Tests: 109/109.
  • ForwardDiff Overloads: 128/128.
  • Traits: 6/6.
  • All later groups passed, ending with SpecializingFactorizations: 18/18.
  • Terminal output: Testing LinearSolve tests passed.

• Whole-repository Runic check and git diff --check both exited 0.

[ChrisRackauckas-Claude]

The downstream ModelingToolkit Julia 1.10 initialization failures have now been split precisely:

• 2 collector-invalidation errors are fixed in Fix collect_vars! after late invalidation on Julia 1.10 ModelingToolkit.jl#4852. The regression is bisected, covered on Julia 1.10/1.12, and the official Julia 1.10 InterfaceI group passes locally (1498 passed, 5 existing broken).
• the remaining 9 errors are the independent DelayDiffEq DEOptions resolver incompatibility tracked by Retroactively cap DelayDiffEq at OrdinaryDiffEqCore 4.8.0 JuliaRegistries/General#162781.

#4852 is a draft pending CI/review; this link is classification, not a claim that this PR's full matrix is green yet.