doesn't this just mean the spec is overwritten? (and covering things that are not in use by the dominant engines)

It's useless to get a higher score on compliance than the leading engines because ... no one else can use them.

The specs and the test suite are both moving targets. There are regularly new proposals to the specs, and new tests that cover them as they progress towards acceptance. The main engines implement these proposals behind feature flags, and only enable them once the proposal has been fully accepted.

Ladybird does not hide implementations behind feature flags (yet) because there's no need when you don't have users. So its score on test262.fyi includes all proposals it has implemented thus far.

The other engines on that site have an "experimental options" variant to include these proposals, which is a bit more of an honest comparison. As of right now, that shows: Spidermonkey (Firefox) at 98.3%, V8 (Chrome) at 97.9%, LibJS (Ladybird) at 96.9%, and JavaScriptCore (Safari) at 93.2%.

Here's a link with those options selected: https://test262.fyi/#|v8_exp,jsc_exp,sm_exp,libjs