[Yuchao Yan (msyyc)]

Bounded source repair for #5526

The original source migration #5523 is merged. This separately authorized follow-up fixes its concrete inherited-operation boundary without reopening that PR or mutating the original reviewed source pin 08ea9770a0209d7a131c467ec0fd93f3918b1714. The source package exists on the owning development queue, so this repair targets feature/lintdiff-migration-new, not main. No merge is requested or performed automatically.

isTemplateInstance(operation) also matches real PUT endpoints inherited from generic interfaces into concrete interfaces. The canonical violating regression previously produced zero warnings instead of one. Use the compiler's declaration-aware predicate, retain genuine template exclusions, and add inherited positive/negative controls with an exact authored-property diagnostic target. Rule identity, message, enablement and authored-vs-JSON-name semantics are unchanged. Source migration guidance and its unreleased fix note match this boundary.

Validation and review

• Dependency-closure and production builds passed at compiler a6137cac43a727ce0c2656364fd72d50c272ee4a.
• Red regression: expected one warning, found zero; the other eleven native cases passed.
• Repaired native 12/12 and independent emission 2/2 passed; strict changed-test types passed.
• Scoped normal/type-aware lint, all five intended-file formatting checks and diff hygiene passed.
• Independent source-development review found no significant issues in frozen tree 0ac45147f27cf61f439bcd3778eafb6e4b7d08f3; the committed tree is identical. The reviewer independently reran all 14 focused tests. This is not GitHub approval.

Root lint is not green: it reports 228 warnings entirely in unchanged harness/other-rule files, zero in the changed source/test. Root formatting rewrote unrelated baseline fixture/metadata files; only formatter-owned spillover was restored, and the final diff is exactly five intended files. Narrow checks follow the repository's generated-artifact/baseline exception. Historical corpus totals were not refreshed and are not claimed as repaired-head evidence. No universal Swagger parity or broad integration/merge readiness is claimed.