fix(tests): unbreak class-id audit (#11691) and sloppy-this gap oracle (#11693)

changelog.d/11694-fix-main-reds-class-id-sloppy-this.md

@@ -0,0 +1 @@
+Fix two reds on `main`, both test-only. (1) Renamed the test-local `CLASS_ID` in `perry-runtime/src/object/static_shapes_tests.rs` (from #11678) to `LATE_POOL_CLASS_ID`. `scripts/class_id_collisions.py` no longer reports it as a drifted mirror of `class_registry/state.rs`'s `CLASS_ID`, and the audit itself is unchanged (#11691). (2) Moved `test_gap_sloppy_this_bound_once` from `.ts` to `.cts`. Under the repo's `"type": "module"`, Node ran the `.ts` as strict ESM and the oracle threw on `mutate.call(1)`. A `.cts` is sloppy CommonJS in both runtimes, and Perry's output now matches Node 26.5.1 byte for byte (#11693).

crates/perry-runtime/src/object/static_shapes_tests.rs

@@ -222,12 +222,15 @@ fn a_seeded_literal_shape_answers_the_megamorphic_confirm_like_a_minted_one() {
 /// class's confirms fail.
 #[test]
 fn a_class_registered_before_the_pools_answers_the_megamorphic_confirm() {
-    const CLASS_ID: u32 = 0x0074_1c77;
+    const LATE_POOL_CLASS_ID: u32 = 0x0074_1c77;
     let packed = b"ltca_x\0ltca_only_in_b\0";
-    let keys =
-        crate::object::js_build_class_keys_array(CLASS_ID, 2, packed.as_ptr(), packed.len() as u32)
-            as u64;
-    let id = shapes::js_object_shape_id_for_class_keys(keys, 2, CLASS_ID);
+    let keys = crate::object::js_build_class_keys_array(
+        LATE_POOL_CLASS_ID,
+        2,
+        packed.as_ptr(),
+        packed.len() as u32,
+    ) as u64;
+    let id = shapes::js_object_shape_id_for_class_keys(keys, 2, LATE_POOL_CLASS_ID);
     // Module B's pool runs afterwards and mints its key literals.
     let (x, b) = (pool_atom("ltca_x"), pool_atom("ltca_only_in_b"));
     assert!(