Open
Description
This issue is a follow-up to #1001, which tracks the effort to decouple the core EPP logic from the InferenceModel resource. Once that refactoring is complete, the conformance tests should be updated to take advantage of the new, more modular architecture.
The GatewayFollowingEPPRouting conformance test, introduced in PR #961, currently requires the creation of an InferenceModel resource to run ( source). This is an implementation detail of the testing EPP, which needs the resource to discover the full list of potential backend pods and identify the model it is serving
As discussed in the PR, the ideal state for this test is to not depend on the InferenceModel at all.