test(studio): the backup connection tests find the passphrase field by the change action it now names, and the Markdown poll test waits out the 220ms tab-entry timer before it takes its baseline

desktop/frontend-next/src/ui/Backup.connection.interaction.test.tsx

@@ -47,7 +47,9 @@ function makePort(...ids: string[]) {
 }
 
 function control<T extends HTMLElement = HTMLButtonElement>(action: string, target?: string) {
-  return document.querySelector<T>(`[data-action="backup.${action}"]${target ? `[data-target="${target}"]` : ""}`)!;
+  const at = target ? `[data-target="${target}"]` : "";
+  // A field that also reacts to a key names each event, so its action sits on the change attribute.
+  return document.querySelector<T>(`[data-action="backup.${action}"]${at}, [data-action-change="backup.${action}"]${at}`)!;
 }
 
 function change(action: string, value: string, target?: string) {

desktop/frontend-next/src/ui/useWorkbenchFiles.poll.test.tsx

@@ -98,8 +98,9 @@ it("leaves the rendered Workbench untouched when Markdown polls are unchanged",
   await user.click(await screen.findByRole("button", { name: "notes.md" }));
   await screen.findByRole("heading", { name: "Saved" });
   await act(async () => {});
-  // Whatever the open itself left pending lands before the baseline, not in it.
-  await act(async () => { await new Promise((resolve) => setTimeout(resolve, 0)); });
+  // A tab that has just opened clears its entering state on a real 220ms timer
+  // (the fake clock covers intervals only), so the baseline waits it out.
+  await act(async () => { await new Promise((resolve) => setTimeout(resolve, 300)); });
   commits.mockClear();
   await tick();
   await tick();