API discovery currently polls all models repeatedly and indefinitely. This will be problematic for our upcoming release as the process is slow, intensive, and expensive (charged per API request per user).

We can improve on this implementation in the short term. Ultimately we will need a backend solution closer to the source of the problem to resolve the issue permanently.

This story tracks the effort to update the SDK to default to the plural of a model (to lower the number of requests we'll ultimately need to make)

AC:

• Update API discovery such that if we don’t have the plural for a model and you request to watch it (through a hook), we will turn around and fetch the plural and perform the watch on that single endpoint (as KCP will likely be structured) and perform API discovery on that model

See 22 November 2022 entry in HAC Core Arch sync doc for more details (slim API discovery): https://docs.google.com/document/d/1v5abxm5Gz6bDGLsdYwZc5ixZ2GiIja7xFt97LoBu7AU