Feature Request]: "Behavior" (BDD) rendering mode for the Interactions panel.

[paul-hammant]

Is your feature request related to a problem? Please describe.

The Interactions panel renders the instrumenter call stream in implementation voice — literal calls with resolved arguments. Two consequences:

1. It documents how the test is written, not what the component promises, so a story
   catalogue can't double as a living spec for anyone who doesn't read testing-library.

2. Printing resolved values loses the subject. A real example from our catalogue
   (a table's Loading story):

   expect(<table.cds--skeleton.cds--data-table>).toBeTruthy()
   expect(null).toBeNull()
   

   That second line was expect(canvas.queryByText('John Doe')).toBeNull() — the one
   fact a reader needs ("John Doe is not shown while loading") is exactly the part
   that evaporated.

Everything needed to fix this already exists: the instrumenter events carry method,
path refs, args, and pass/fail status; story metadata carries the title path and story
name. The gap is purely presentational.

Describe the solution you'd like

An optional rendering mode on the Interactions panel (toggle or parameter) that
renders the same event stream as behavior sentences:

BDD rung	Source
nested describe	title path split on / (leaf = component)
context / Given	story name
when …	userEvent.* / fireEvent.* calls
<subject> should …	expect matcher calls, matcher name humanized (toHaveClass → "should have class")

Locator calls (within, getBy*, queryBy*, findBy*) and bare expect(target)
calls are suppressed from the log but mined for the subject (e.g. the string arg of
getByLabelText('Admin')), threaded onto the matcher line. The two lines above become:

✓ should be truthy
✓ John Doe should be null

and an interaction test reads:

✓ when click Save
✓ Label is required should be in the document
✓ Fields is required should be in the document
✓ onSave should not have been called

Doing this inside Storybook is strictly better than userland: the instrumenter still
has expression context at record time, whereas an external consumer must reconstruct
subjects from __callId__ refs and irrecoverably loses them whenever a query resolves
to null or a primitive (the expect(null) case above).

Describe alternatives you've considered

• Userland overlay (what we run today): we inject a script via the test-runner's
  prepare hook that subscribes to storybook/instrumenter/call and renders a live
  BDD log into the preview during headed test-runner passes (~200 lines). It works,
  but it couples to undocumented channel internals, has the null-subject limitation
  above, and we deliberately don't ship it to storybook build output — which is
  exactly where non-developer readers are.
  • Post-processing test-runner output: loses live/interactive value and the
    panel's step-through affordances.

Are you able to assist to bring the feature to reality?

yes, I can

Additional context

Observed with Storybook 10.3.6 (@storybook/react-webpack5), @storybook/test-runner
0.24.4, instrumenter active (instrument=true). The value proposition in one line:
the Interactions panel currently answers "what did the test do?"; a behavior mode
would answer "what does the component promise?" — turning every story catalogue into
living documentation for free.