Random failure in component-operations.cy.js

Needs Cypress expertise to review, due to the introduction of waits and an intercept, which I thought we eliminated? 🤔

I believe this is similar to my comment in #3500136: Random failure in empty-canvas.cy.js where there are waits being added for endpoints to address the issue, but the problem occurs slightly earlier so those endpoints are not being requested at all.

This seems to be reliably passing with the changes I made in #3500542: Implement adding component to empty layout without dnd. It's difficult to be 100% sure since it is intermittent but I re-ran the tests many times and it's green every time

#3500542: Implement adding component to empty layout without dnd has been merged.

But … #9: I found a remaining random failure, that is not related to drag-and-drop:

FYI: when this was merged, component-operations.cy.js failed during the 0.x test run 🙈 But it failed ~20 LoC earlier than the change here.

Failure:

      cy:command ✔  click	
        cy:fetch ➟  GET http://localhost/web/xb/api/config/pattern
                    Status: 200
      cy:command ✔  get	.primaryPanelContent
      cy:command ✔  within	
      cy:command ✔  findByText	Sections
      cy:command ✔  click	{force: true, scrollbehavior: false}
      cy:command ✘  findByText	The Entire Node 1
      cy:command ✘  assert	expected **findByText(`The Entire Node 1`)** to exist in the DOM
                    Actual: 	"findByText(`The Entire Node 1`)"
                    Expected: 	"findByText(`The Entire Node 1`)"

Relevant test code:

    cy.openLibraryPanel();
    cy.get('.primaryPanelContent').within(() => {
      cy.findByText('Sections').click(clickDefault);
      cy.findByText('The Entire Node 1').should('exist');
    });
    cy.get('.primaryPanelContent')
      .as('panel')
      .should('contain.text', 'The Entire Node 1');

— #3500542-9: Implement adding component to empty layout without dnd

component-operations had two different random fail areas, one of which was recently addressed in #3500542: Implement adding component to empty layout without dnd. The other, which happens to be the screenshot in the issue summary, has something pretty specific going on.

It fails when looking for a section called "The Entire Node 1". In the issue summary, that section is instead namede "Node 1"

In a separate instance of this failing, the section is named "e Node 1"

This is the code where it happens - the input is receptive enough to receive the clear() as there are no artifacts of the default section name "Two Column section", but it looks like the fails occur when some of the initial typing used to name the section fails to register.

   cy.findByLabelText('Section name').clear();
    cy.findByLabelText('Section name').type('The Entire Node 1');
    cy.findByText('Add to Library').click({ scrollBehavior: false });

It is a pretty specific way to fail so hopefully that means a fix is reasonably easy to achieve too.

Fortunately I was able to reproduce the flaky test locally by slowing things down a bit (throttling or enabling Xdebug worked). The 3500128-retry-typing-until-it-works branch has the solution, and I've re-run it several times. It being intermittent makes it hard to be 100% certain, but this fixing it on a local where it happened reliably