fix: code suggestions uses v3 endpoint for code generations requests

What does this merge request do and why?

This MR addresses gitlab-org/ai-powered/eli5#67 (closed)

Eli5 currently encounters a validation error when making code generation requests to the /v2/code/generations endpoint in AIGW during model evaluations. This occurs because the endpoint requires a model_endpoint parameter.

This issue aims to resolve this problem by updating Eli5 to use the /v3/code/completions endpoint instead for code generations requests, which doesn't require the model_endpoint parameter. This change also aligns Eli5's code suggestion functionality with production usage patterns and allows us to evaluate and compare model performances for our datasets.

Related to: gitlab-org/gitlab#508167 (closed)

How to set up and validate locally

Confirm code generation requests work with the new endpoint and a model provider, when no provider is given, it defaults to vertex-ai.

poetry run eli5 code-suggestions evaluate \
  --dataset="dataset.code_suggestions_generations_squadri.6" \
  --source=ai_gateway \
  --limit=1 \
  --experiment-prefix=shola-claude-3-codegen-exp \
  --intent=generation \
  --model-provider="anthropic" \
  --evaluate-with-llm

Confirm code completions requests function as before:

poetry run eli5 code-suggestions evaluate \
--dataset="code-suggestions-input-testcases-v1" \
--source=ai_gateway \
--evaluate-with-llm \
--limit=1 \
--experiment-prefix=experiment-v3-endpoint

Numbered steps to set up and validate the change are strongly suggested.

Merge request checklist

• I've ran the affected pipeline(s) to validate that nothing is broken.
• Tests added for new functionality. If not, please raise an issue to follow up.
• Documentation added/updated, if needed.