Provide web API for testing AI availability

Problem

This is a follow-up and improvement to #466698 (closed). It will directly support the UX defined in #454970 (closed).

In order to improve the AI onboarding experience, we should provider a web API (GraphQL or REST) that the front-end can use to check in with the backend to signal the user whether AI features are available and working.

Requirements

• Should cover AI broadly not just Code Suggestions
• Should be designed so that it will be easy to extend to non-AI features in the future

Stretch goals

• Should be an asynchronous call so that we don't block page rendering on it (for example, we may have to wait several seconds for cloud.gitlab.com to respond, which would delay page rendering by the same amount)

TODO

We should define design and UX for this first, then model the API to fit.