test(cutover): the ui lazy-node pins count five since spec's shared chunks (bill item 37, objectstack#25803)

Cutover PR objectstack#25825 run 1, Test UI 14/16 and 15/16: imported-defaults-8317 expects 6
reachable `z.lazy` nodes and finds 5, and zod-mirror-parity's UNNAMED_LAZY_SLOTS still lists
`form.zod.ts#FormFieldSchema.summaryOperations`, which the live walk no longer reports. No schema
changed (FilterConditionSchema is byte-identical between the pin 5e7cf63 and main): spec
objectstack#25748 ships one module graph with shared chunks, so each schema is one object across
entries, and the lazy reached through `FieldSchema.summaryOperations.filter` is now the same
object as the recursive filter clause (measured: 5 distinct lazies, that path in the filter
clause's group). Both tests are ratchets built to red on exactly this event.

The pin moves 6 -> 5 with the reason, the summaryOperations row goes (its comment stays as
history), and imported-defaults.ts's walker docblock re-derives its count as five.

Measured on the cutover head 1c0d02e + the series: 2 failed / 312 passed -> 314/314.

Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CUXv64KLMfYv3ia2sWi6te

Author: claude

ui/packages/types/src/__tests__/imported-defaults-8317.test.ts

@@ -492,17 +492,18 @@ describe('the import boundary strips every imported default (objectui#8317)', ()
     it('the walker docblock\'s `lazy` count is re-derived, not quoted', () => {
       // The `lazy` arm is the one place the identity property cannot hold: it
       // must rebuild without forcing the getter, so a clean subtree behind a
-      // `z.lazy` is rebuilt anyway. The module's docblock names SIX such
+      // `z.lazy` is rebuilt anyway. The module's docblock names FIVE such
       // nodes (the fourth, `ViewSchema`'s form-field group, since objectui#7928;
-      // the fifth, under `FieldSchema.relatedListFilter`, since objectui#11070;
-      // the sixth, the form-field group under the `object-form` row's
-      // `sections[].fields[]`, since the 17.7.0 bump, objectui#11717)
+      // the one under `FieldSchema`, added by objectui#11070, is the recursive
+      // filter clause's own object since objectstack#25748's shared chunks; the
+      // form-field group under the `object-form` row's `sections[].fields[]`,
+      // since the 17.7.0 bump, objectui#11717)
       // and says the exception costs nothing extra for every owner but the
       // ones `REBUILT_CLEAN` names (objectui#10872 batch 2), because each
       // other owner sits inside a schema that is being rebuilt regardless.
       // Both halves are measured here, so a spec bump or an import that moves
       // either one is red rather than quietly making the docblock false.
-      expect(walk(IMPORTED.map(([, s]) => s)).lazies).toBe(6);
+      expect(walk(IMPORTED.map(([, s]) => s)).lazies).toBe(5);
       const lazyOwners = IMPORTED.filter(([, s]) => walk([s]).lazies > 0);
       expect(lazyOwners.length, 'no schema owns a lazy — the count above found them elsewhere').toBeGreaterThan(0);
       // The clean owners are exactly the named set: rebuilt (so NOT the spec's

ui/packages/types/src/__tests__/zod-mirror-parity.test.ts

@@ -6353,7 +6353,9 @@ const UNNAMED_LAZY_SLOTS: readonly string[] = [
   // `FieldSchema` member by reference, and its optional `filter` is the spec's
   // recursive filter condition. That `z.lazy` sits inside an IMPORTED schema
   // (rebuilt by `stripImportedDefaults`), so no local const names it either.
-  'form.zod.ts#FormFieldSchema.summaryOperations',
+  // Retired by objectstack#25748: the spec's shared chunks make that lazy the
+  // same object as a named recursive mirror reached through another entry, so
+  // the walk no longer reports it as unnamed (the schema itself is unchanged).
   'objectql.zod.ts#ObjectViewSchema.form',
   'objectql.zod.ts#ObjectViewSchema.table',
 ];

ui/packages/types/src/zod/imported-defaults.ts

@@ -141,6 +141,12 @@ const walk = (schema: z.ZodType): z.ZodType => {
   // recursive form-field group under `ObjectFormPropsSchema.sections[].fields[]`,
   // reached from the `object-form` and `object-master-detail-form` rows, whose
   // `sections` take the page-block section shape since that release.
+  // objectstack#25748 then made the count FIVE: `@objectstack/spec` ships one
+  // module graph with shared chunks, so each schema is one object across its
+  // entries, and the lazy `FieldSchema` reaches (walked today as
+  // `FieldSchema.summaryOperations.filter`) is now the SAME object as the
+  // recursive filter clause above rather than a copy of it. The fifth is no
+  // longer a node of its own; nothing in the schemas changed.
   // Each sits inside a schema that carries a default anyway, so the exception
   // costs no extra rebuild. objectui#10872 batch 2 had ONE named exception,
   // `ElementDataSourceSchema` (the `element:number` arm's `dataSource`), whose