We are skipping opt-in related e2e tests since this involves user-based storage in Atlas and can get complicated to mock. We'd probably want to investigate how to change the AI Assistant features override dynamically with tests. The new `browser.setEnv` function should help but gets complicated with webpack replacing process.env variables; We might want to define a custom environment variable which we preserve in runtime. 

We also skip NLQ tests completely a the moment so we'd want to use the opportunity to enable those too.