docs(changelog): clarify MethodToMethodWithCheckRector BC-wrap edge case

The bullet under [Unreleased] / ### Changed previously claimed the BC
wrap "addresses the underlying concern" of the dropped maybe-path TODO
comment in all cases. That's only true for configurations with
introducedVersion >= 10.0.0 — pre-10.0 entries (urlInfo, getLowercaseLabel,
clearCsrfTokenSeed) fall outside the BC-wrap gate and rewrite a
maybe-typed receiver unconditionally.

Contrib audit (api.tresbien.tech) found zero live callers of those three
methods on receivers PHPStan cannot resolve, so the residual risk is
theoretical. Updates the CHANGELOG bullet to acknowledge this explicitly
rather than overclaim safety.

Author: bbrala

CHANGELOG.md

@@ -396,9 +396,16 @@ release-by-release.
 - `MethodToMethodWithCheckRector` no longer attaches a "please confirm the
   receiver type" TODO comment for the `maybe` type-inference case. The
   comment-on-parent-statement mechanism relied on parent-node tracking that
-  Rector 2.x removed; the BC wrap makes both code paths runtime-safe by
-  selecting the right call based on `\Drupal::VERSION`, which addresses the
-  underlying concern.
+  Rector 2.x removed. For configurations whose `introducedVersion ≥ 10.0.0`,
+  the BC wrap selects the right call at runtime via `\Drupal::VERSION` and
+  fully addresses the underlying concern. Configurations with an older
+  introduced version (e.g. `urlInfo` → `toUrl` @ 8.0.0,
+  `getLowercaseLabel` → `getSingularLabel` @ 8.8.0,
+  `clearCsrfTokenSeed` → `stampNew` @ 9.2.0) fall outside the BC-wrap gate
+  and rewrite unconditionally on a `maybe`-typed receiver. Contrib audit
+  (api.tresbien.tech, 2026-05-28) found zero live callers of those three
+  methods on receivers PHPStan cannot resolve to a concrete type, so the
+  residual risk is theoretical.
 
 ## [1.0.0-beta1] — 2026-05-25