Add an early-return path for static virtual methods matching the C++ implementation in MethodTable::TryResolveConstraintMethodApprox. The C++ handles static virtuals separately by calling ResolveVirtualStaticMethod, and only does so when neither the interface method nor the interface type are shared by generic instantiations.

The C# now mirrors this: when the method, interface type, or constrained type involves canonical forms, resolution is skipped. Otherwise, the method is resolved via ResolveVariantInterfaceMethodToStaticVirtualMethodOnType with fallback to ResolveVariantInterfaceMethodToDefaultImplementationOnType.

This fixes a pre-existing bug where variant static virtual default interface method dispatch was not properly resolved in the NativeAOT compiler.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

copilot --resume=98a1ac45-917c-4269-bff4-fb71fa3c8f35

Rename local variables in FindDefaultInterfaceImplementation and
TryGetCandidateImplementation to match the C++ implementation:
- runtimeInterface -> currentMT (pCurMT)
- interfaceMethodOwningType -> interfaceMT (pInterfaceMT)
- candidateMethod -> candidateMD/currentMD (candidateMD/pCurMD)
- bestCandidateInterface -> bestCandidateMT (pBestCandidateMT)
- bestCandidateMethod -> bestCandidateMD (pBestCandidateMD)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

… check

Replace the over-conservative constrainedType.IsCanonicalSubtype guard with
the canonicalEquivalentFound check from C++ ResolveVirtualStaticMethod. This
matches the C++ behavior: only bail out when the canonical form of the
interface is also in the constrained type's interface map (indicating
ambiguous resolution in shared generic code).

Relax Scanner and RyuJIT assertions to allow static method results on
non-value types, since constrained static virtual calls can resolve to
methods on reference types (e.g. GenericClass<T> implementing an interface).

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Move the static virtual method check before the value type check,
matching the C++ ordering in MethodTable::TryResolveConstraintMethodApprox:
1. if (IsStatic()) - handle static virtuals first, return early
2. if (!IsValueType()) - then reject non-value-types for instance methods

This removes the IsCanonicalDefinitionType carve-out from the value type
guard since static virtuals are now fully handled before it runs.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Since static virtual methods now return early at the top of
TryResolveConstraintMethodApprox, the remaining isStaticVirtualMethod
checks in the instance method path are unreachable dead code. Remove
them and inline the check at the entry point.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Add four new test cases for static virtual method dispatch through
shared generic constrained calls:
- StaticVirtualOnReferenceType: class-based dispatch
- GenericStaticVirtualMethod: method-level generic type parameters
- CanonicalEquivalentFound: canonical equivalence in shared code
- StaticVirtualDIMOnReferenceType: default interface methods on classes

Fix ILC Scanner crash for GVM-only interfaces by swapping the check
order in ConstrainedMethodUseLookupResult.NonRelocDependenciesFromUsage.
The HasInstantiation check is now evaluated before factory.VTable() to
avoid calling ScannedVTableProvider.GetSlice for interfaces whose VTable
was never scanned (because GVMs are excluded from VTable slots and
VirtualMethodUse is not reported for them).

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…on NativeAOT

- Add third pass in DispatchResolve for variant default interface method dispatch
- Filter inherited DIM entries in InterfaceDispatchMapNode to avoid redundant
  dispatch map entries that would cause incorrect resolution order
- Add VariantDIMIterationOrder regression test
- Remove ActiveIssue for NativeAOT on ComplexHierarchyPositive (dotnet#88690) and add
  dynamic dispatch tests via reflection

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>