Steps to reproduce
0. Run Ladybird and notice the messages logged in the terminal

1. Go to http://dagbladet.no (newspaper)
2. Once loaded, it will continue to print the following messages:

ResourceLoader: Starting load of: "https://www.dagbladet.no/app/cyclops-router-api/api/v1/users/me"
ResourceLoader: Failed load of: "https://www.dagbladet.no/app/cyclops-router-api/api/v1/users/me", Error: Load failed: 401, Duration: 61ms
Fetch: Username/password prompt is not implemented, using empty strings. This request will probably fail.

(I assume it retries because it failed and it fails due to missing implementation. That's fair enough, and will probably be resolved at some point ™️ )
3. Go to another site, for instance http://github.com
4. Notice that even though you are at a different page, the log messages continue.

Expected:
Once I navigate away from a website, any requests from that context will be stopped.

Actual:
The request attempts linger and keep looping until I either close the browser or at least the tab where they originated. I don't know how, but it looks like processes like this can survive navigation to another domain and still run in the background.

(I'm not sure how to trim this down to a minimal test case, but the behaviour seems to be reproducible for that site.)