fix(openai): add structured output instrumentation

[9dogs]

• I have added tests that cover my changes.
• If adding a new instrumentation or changing an existing one, I've added screenshots from some observability platform showing the change.
• PR name follows conventional commits format: feat(instrumentation): ... or fix(instrumentation): ....
• (If applicable) I have updated the documentation accordingly.

Fixes #1815. I'm not sure if the response_format span attribute should be added in this PR as well.

Also, the metrics tests are a bit fragile since each executed test adds its metrics to the reader.get_metrics_data() result, so figuring out what was added is a bit tricky.

[CLAassistant]

All committers have signed the CLA.

[nirga]

Thanks @9dogs! We add the response format here or in a separate PR - up to you (if you do that - we might as well add everything from the recently published semantic conventions)

[9dogs]

Thanks @9dogs! We add the response format here or in a separate PR - up to you (if you do that - we might as well add everything from the recently published semantic conventions)

I cannot say I fully understand what's going on with semconvs, so let's leave the scope of this PR as it is.

[nirga]

Thanks @9dogs! Left a small comment

[nirga]

@9dogs we should wrap these in try-except since this may fail on old OpenAI SDK versions (this is why the tests are currently failing)

[nirga]

Thanks @9dogs!