LibWeb currently implements a process-per-tab architecture.

To improve stability and security, we should move to a process-per-origin architecture.

This means that we'd spawn additional WebContent processes for each unrelated origin that participates in a page, and then proxy all access to it over IPC as needed.

Note that this also means that iframes within a page may have their content rendered by another process, so we'll need to improve our painting architecture quite a bit here.