[Vertex AI] Add integration tests for staging endpoint

[andrewheard]

Updated CountTokensIntegrationTests and GenerateContentIntegrationTests to test against the staging endpoint in addition to the existing prod tests. This will allow us to catch any regressions before backend builds reach prod.

#no-changelog

[gemini-code-assist]

Code Review

This pull request introduces integration tests for the staging endpoint in addition to the existing prod tests. This is a good approach to catch regressions before backend builds reach prod. The changes look good overall, but I have a few suggestions for improvement.

Summary of Findings

• APIConfig enum case: The vertexAI case in the APIConfig enum now takes an endpoint parameter. This is a good change to support different endpoints. However, it might be beneficial to consider a more descriptive name for the enum case to improve clarity.
• Assertions in VertexAI.swift: The assertions in the VertexAI.swift file now explicitly check for the .vertexAI(endpoint: .firebaseVertexAIProd) endpoint. While this is correct, it might be beneficial to add a comment explaining why these assertions are necessary and what they are verifying.
• InstanceConfig definitions: The InstanceConfig struct in InstanceConfig.swift now includes definitions for both prod and staging endpoints. This is a good approach for testing against different environments. However, it might be beneficial to add documentation or comments explaining the purpose of each configuration.

Merge Readiness

The pull request is well-structured and introduces a valuable addition to the testing suite. I recommend addressing the comments provided to enhance code clarity and maintainability. Once these suggestions are incorporated, the pull request will be ready for merging. I am unable to directly approve this pull request, and recommend that others review and approve this code before merging.

[andrewheard]

cc: @rainshen49