[OpenAI] "Basic LLM Test" failures: Missing `gen_ai.usage.input_tokens` and `gen_ai.usage.output_tokens` in response

[ericapisani]

Within the "Basic LLM Test" test suite from testing-ai-integrations,there are a couple of failing tests due to these properties not being returned in the response. Determine why that is and fix.

[linear-code]

PY-2271