feat(check): report a feature-owned multiplicity's featuring types and record type-system validation status

[devin-ai-integration]

What and why

docs/project/spec-compliance.md listed four Type System items as not implemented — "Full generic/specialization validation", "Interface realization", "Redefinition conformance checking", "Subsetting validation". An audit of each against the pilot's named validate* constraints found the list stale: every named specialization, redefinition and subsetting constraint is already reported (✅ in the census), one KerML rule was genuinely missing, and "interface realization" is not a SysML v2 concept.

• Implemented validateFeatureMultiplicityDomain (KerML 1.0 §8.3.3.3.4): a feature-owned multiplicity must have the same featuring types as its feature. The only textual way to violate it is a standalone featuring of C::x::m by D; on the multiplicity, which MultiplicityDomainPass (constraint tier, error, pilot wording Multiplicity must have same featuring types as it feature) now reports on the relationship member. A target that is one of the feature's featuring contexts, or specializes one, is accepted; classifier-owned multiplicities (validateClassifierMultiplicityDomain, ⛔ deliberate) and unresolved ends are left alone. Census row ❌ → ✅ with a probe; the pipeline self-model gains the pass (56 → 57).
• The six ⚠️ rows stay approximate, each now stating why in validation-constraints.md: validateBindingConnectorTypeConformance (the value/feature typing judgment is the type tier's, where a mismatch is an error — a constraint-tier warning would let the constraint tier run over an ill-typed model), validateClassifierDefaultSupertype (the Links::BinaryLink requirement is KerML §8.3.4.7 and is kept), validateFlowEndImplicitSubsetting (the same end is already rejected as inaccessible; the pilot's warning beside it would report one defect twice), validateExposeOwningNamespace and validateRequirementConstraintMembershipOwningType (deliberate extensions: warning by default, -strict matches the pilot), validateTransitionUsageSuccession (the state-machine reading is what the state executor runs).
• Docs: the Type System bullets are rewritten with status, file:function and test per rule; "Interface realization" is removed (UML InterfaceRealization has no v2 counterpart — an interface def is a connection between ports, and the two sides of an interaction are port conjugation); roadmap Track V figures updated; census counts regenerated.

Audit summary (rule → pilot constraint → status):

Bullet	Pilot constraints	Status
Specialization	validate{Class,DataType,Structure,Behavior}Specialization, validate{Association,Connector}BinarySpecialization, validateSpecializationSpecificNotConjugated, validateTypeAtMostOneConjugator, validateFeatureHasType, validate*UsageType, validate{Definition,Usage}VariationSpecialization	✅; validateClassifierDefaultSupertype ⚠️ (stricter, deliberate). "Generic" types do not exist in KerML/SysML v2
Interface realization	—	not a SysML v2 concept (removed); the v2 rules that do exist — validateInterface{Definition,Usage}End, validateInterfaceUsageType, validatePortUsage* — are ✅
Redefinition conformance	validateRedefinition{FeaturingTypes,EndConformance,DirectionConformance,MultiplicityConformance}, validateSubsetting{Constant,Uniqueness}Conformance, validateFeatureValueOverriding	✅ (multiplicity as a warning, like the pilot)
Subsetting	validateSubsetting{FeaturingTypes,MultiplicityConformance}, validateFeatureOwned{Reference,Cross}Subsetting, validateCrossSubsetting*, validateFeatureCrossFeature*, validateFlowEndSubsetting, validateFeatureMultiplicityDomain	✅ (the last one landed here); validateFlowEndImplicitSubsetting ⚠️; validateSubsettingPortionConformance is a no-op in the pinned pilot

Census: 162 → 163 of 217 reported; 156 → 157 ✅; 6 ⚠️; 1 → 0 ❌; 1 ⛔; 0 🚧; 53 ❔.

Specification basis

KerML 1.0 §8.3.3.3.4 (Feature: multiplicity.featuringType = featuringType), the pilot's KerMLValidator.validateFeatureMultiplicityDomain. Moves the census row in docs/project/validation-constraints.md and rewrites the Type System entries in docs/project/spec-compliance.md.

How it was verified

• internal/check/passes/multiplicity_domain_test.go: package-level and body-level standalone featuring are reported once each; a target that is the feature's owner, a featuring on the feature itself, a classifier-owned multiplicity and an unresolved target are silent.
• Census probe tools/census/validation/testdata/probes/validateFeatureMultiplicityDomain.kerml; go run -C tools ./cmd/validation-census -check and go test -C tools ./census/validation pass. The pinned pilot KerML validator was run over the probe and both negative shapes by hand: silent on all three (its featuringType reads only owned type featurings; its only trigger for this constraint is the defective alias/reference derivation recorded in omg-issues.md, which stays a valid model here).
• False positives: featuring occurs in the corpora only in kerml-examples/Simple Tests/Features.kerml and two rejection fixtures; a sweep over examples/, tests/testdata/ and the fixtures reports zero multiplicity-featuring-type diagnostics.
• Pilot referees, before and after, identical structured totals (jar not found: 0):
  • differential: 378 file(s), 347 fully agreeing; 38 agreed diagnostic(s), 38 only ours, 1185 only the pilot's
  • xpect: 429 .xt file(s), 0 unparsed; 1264 assertion(s), 1326 expectation(s): 1296 agree (of which 248 wording-only), 30 disagree, 0 unlocated, 0 not adjudicated
  • rejection: 306 case(s): 297 both reject, 0 only the pilot rejects, 9 only we reject, 0 both accept
  • The one baseline movement is pilot-differential-baseline.json provenance.inputs[examples].digest: the self-model gained the new pass; no diagnostic or total moved. No rejection case was added — the pilot accepts the model, which the rejection corpus's policy records as a coverage gap rather than a case. training_examples_expected.txt is untouched.
• Gates (all exit 0):

  go build ./... && go vet ./... && gofmt -l .        # gofmt prints nothing
  OPENSYSML_REQUIRE_TRAINING_CORPUS=1 OPENSYSML_REQUIRE_PILOT_CORPORA=1 OPENSYSML_REQUIRE_PILOT_LIBRARY_XMI=1 go test ./...   # ok, incl. tests/corpus, tests/identity
  go test -C tools ./...                               # ok
  make docs-check                                      # 0 broken links; no internal labels; census/suite figure tests OK
  python3 scripts/changelog.py check                   # ok
  

Checklist

• make test and make lint pass locally
• Tests added or updated for the change
• Documentation extended where it already covers the surface (see CONTRIBUTING.md)
• Changelog entry added as changes/unreleased/<slug>.<section>.md, not as an edit to CHANGELOG.md
• baselines regenerated and make docs-counts run if a gate count moved (compliance rows need nothing: the census is counted at docs build)
• No internal work-item labels (waves, slices, F4, K5) in the body, docs, or changelog

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring

[devin-ai-integration]

Devin Review found 1 potential issue.