refactor(resources): correct the row-cache premise and clear an R1b false positive

Two things the migration notes got wrong, now fixed in the places that assert them.

The playbook claimed an open table and an embedded one "should hit one cache
entry, and forking that doubles every fetch". Half right, and the wrong half was
driving a planned change. The SCHEMA is keyed on the table id alone, so it
genuinely is one entry on every surface. ROWS are not, and should not be:
`pageSize` is in the key, the page pulls 1000-row pages so `ensureAllRowsLoaded`
can drain for select-all/export in few round trips, and a panel wants a fast
first screen and never drains. One number would either make a chat panel fetch
1000 rows to show ten, or make every bulk operation ten times the requests. The
divergence is correct; both the hook and the playbook now say so.

`MiniTablePanel` -> `LandingLeadsPanel`. It is hardcoded landing-hero markup
with no queries and no relationship to the table resource, so a `Mini` +
resource-noun name misread as a per-consumer fork of the real view — to a reader
and to the audit. R1b baseline drops 12 -> 11.

The R1b note itself was also wrong, claiming every `Embedded*` drops off as its
kind gets a canonical view. The `*Actions` members are tab chrome, not views:
`EmbeddedFileActions` and `EmbeddedInterfaceActions` outlived their migrations,
and `EmbeddedTableActions` just outlived table's. Only the content components go.

Author: waleedlatif1

apps/sim/app/(landing)/components/landing-preview/components/landing-preview-home/landing-preview-home.tsx

@@ -254,7 +254,7 @@ export const LandingPreviewHome = memo(function LandingPreviewHome({
                   animate={{ width: '55%', opacity: 1 }}
                   transition={{ duration: 0.35, ease: EASE_OUT }}
                 >
-                  <MiniTablePanel />
+                  <LandingLeadsPanel />
                 </m.div>
               )}
             </AnimatePresence>
@@ -382,9 +382,14 @@ function ChatMarkdown({
 }
 
 /**
- * Mini Customer Leads table panel matching the resource panel pattern.
+ * The "Customer Leads" panel in the landing hero's static product mock.
+ *
+ * Named for what it is rather than `MiniTablePanel`: it is hardcoded marketing
+ * markup with no queries and no relationship to the table resource, and a
+ * `Mini` + resource-noun name reads to both a reader and the resource-view
+ * audit as a per-consumer fork of the real table view.
  */
-function MiniTablePanel() {
+function LandingLeadsPanel() {
   return (
     <div className='flex h-full w-full flex-col bg-[var(--surface-2)]'>
       <div className='flex items-center gap-2 border-[var(--border)] border-b px-3 py-2'>

apps/sim/components/resources/MIGRATING-RESOURCES.md

@@ -140,10 +140,15 @@ page-size split documented on `TABLE_VIEW_PAGE_SIZE`.
 `tables/[tableId]/`. Nothing renders differently. Land this alone so the diff is reviewable.
 
 **PR 2 — give it the axes.** `TableView` takes `{ source, grants, host }`. Rows and schema resolve
-from the source: workspace scope keeps the existing `tableKeys` factory deliberately — an open table
-and an interface's table module *should* hit one cache entry, and forking that doubles every fetch —
-while share scope reads a `(token, grantId)` route. Replace the interfaces mini-table with a mount
-and delete it.
+from the source: workspace scope keeps the existing `tableKeys` factory, while share scope reads a
+`(token, grantId)` route. Replace the interfaces mini-table with a mount and delete it.
+
+*Corrected while doing it:* this section used to claim the page and an embedded table "should hit one
+cache entry, and forking that doubles every fetch". Half right. The **schema** is keyed on the table
+id alone, so it genuinely is one entry everywhere. **Rows** are not, and should not be: `pageSize` is
+in the key, the page pulls 1000-row pages so `ensureAllRowsLoaded` can drain for select-all/export in
+few round trips, and a panel wants a fast first screen and never drains. One number would either make
+a chat panel fetch 1000 rows to show ten, or make every bulk operation ten times the requests.
 
 **PR 3 — public surface.** Token-scoped rows route, `ResourceSeedMap['table']` carrying the column
 schema, `/t/[token]` or a table module inside a shared interface.

apps/sim/components/resources/table-view/hooks/use-table-source.ts

@@ -16,12 +16,18 @@ import { shareTableSchema } from '@/resources/table-source'
  * Owned here rather than taken from the caller, so two hosts of this view cannot
  * key differently: `pageSize` is part of the infinite-rows query key.
  *
- * KNOWN GAP — this does *not* yet share a cache entry with the tables page,
- * which requests `TABLE_LIMITS.MAX_QUERY_LIMIT` (1000) because its grid drains
- * every row for client-side sort and filter. Opening a table and viewing it in a
- * panel therefore drains it twice. Unifying them is a real behaviour change on
- * one surface or the other — either the page fetches in 100s or a panel fetches
- * 1000 rows up front — so it is deliberately not smuggled into this move.
+ * Deliberately NOT the tables page's size, and the two should not be unified.
+ * `pageSize` is part of the infinite-rows key, so the page and this view keep
+ * separate row caches — which is correct, because they read differently. The
+ * page requests `TABLE_LIMITS.MAX_QUERY_LIMIT` (1000) so `ensureAllRowsLoaded`
+ * can drain a table for select-all, export and bulk delete in few round trips;
+ * a panel wants its first screen fast and never drains. Forcing one number
+ * would either make a chat panel fetch 1000 rows to show ten, or make every
+ * bulk operation ten times as many requests.
+ *
+ * The expensive shared half *is* shared: the schema is keyed on the table id
+ * alone (`tableKeys.detail`), so every surface showing a given table reads one
+ * entry for its columns.
  *
  * A share source uses {@link PUBLIC_TABLE_PAGE_SIZE} instead — the public
  * contract's hard `limit` ceiling, read from the contract rather than restated.

scripts/check-resource-views.ts

@@ -78,13 +78,19 @@ const RESOURCE_POLICY_BASELINE = {
    */
   wrapperMounts: 0,
   /**
-   * R1b — components whose NAME announces a per-consumer fork. 11 of the 12 are
-   * the mothership panel's `Embedded*` tab chrome for kinds that have no
-   * canonical view yet (workflow, folder, scheduled task, log, knowledge base);
-   * the last is a landing-page marketing mock. Every one drops off as its kind
-   * gets a real view.
+   * R1b — components whose NAME announces a per-consumer fork. All 11 are the
+   * mothership panel's `Embedded*` tab chrome.
+   *
+   * These do NOT all drop off as kinds get canonical views, which an earlier
+   * version of this note claimed: the `Embedded*Actions` members are tab chrome
+   * (open / export buttons), not views, and `EmbeddedFileActions` and
+   * `EmbeddedInterfaceActions` both outlived their kinds' migrations —
+   * `EmbeddedTableActions` outlived table's. Only the content components
+   * (`EmbeddedWorkflow`, `EmbeddedFolder`, `EmbeddedScheduledTask`,
+   * `EmbeddedLog`) go. Collapsing the five `*Actions` into one kind-keyed
+   * component is the real fix and is its own change.
    */
-  shadowNamedComponents: 12,
+  shadowNamedComponents: 11,
   /**
    * R2 — imports that reach past a unit barrel. At its floor: the three
    * legitimate deep imports are all `lazy()` code-split points, listed in