The issue you've described sounds like a defect with the API, if all of the assumptions you outlined are correct. From a syntax perspective, your test is written correctly.

Does the API respond with a 200 and a null body, or a different response code?

Are you sure the endpoint that establishes state for your application is the one that is returning null? Providing more context about the implementation of the system under test may point to the issue more clearly.

Have you asked the developers of the application you're testing what the expected behavior is on a session resume? As this is an end-to-end test, it should normally be sufficient to "do what a user would do" and wait the allowed amount of delay for the system to pass assertions without interrogating the implementation (local storage, API payloads, etc). As recommended by Playwright's docs: test user-visible behavior