RawHTML creates a `<div>`, which isn't a valid descendant of `<p>`.

Without this, the user would see search result entries like`[]` instead of `[post]`, until their content index was updated via new posts, etc.

`wpwrap` doesn't exist, so an error would be thrown when visiting the login page and the action was anything other than `loggedOut`.

Blink/WebKit-based browsers don't expose `window.caches` over HTTP, even though they should. Previously those users would get an error that the index can't be fetched, and a request to update to a newer browser. We can still fetch the index even if we can't cache it, though; that's a performance hit, but a much better UX overall.