feat: Add OpenAI Batch API Instrumentation

[Luca Forstner (lforst)]

Adds resumable tracing for OpenAI Batch jobs. openaiFilesCreateTraced() starts a task span with pending LLM spans from a batch input upload; openaiBatchesRetrieveTraced() records OpenAI lifecycle timestamps; completeOpenAIBatchTrace() adds outputs and errors from caller-supplied batch files.

const inputFile = await openaiFilesCreateTraced(client.files)({
  file,
  purpose: "batch",
});

const batch = await client.batches.create({
  input_file_id: inputFile.id,
  endpoint: "/v1/chat/completions",
  completion_window: "24h",
});

Complete the trace from a poller or webhook handler. File requests start concurrently; supplied response bodies are streamed and consumed:

const batch = await openaiBatchesRetrieveTraced(client.batches)(batchId);

await completeOpenAIBatchTrace({
  inputFileId: batch.input_file_id,
  inputFileContent: client.files.content(batch.input_file_id),
  outputFileContent: batch.output_file_id
    ? client.files.content(batch.output_file_id)
    : undefined,
  errorFileContent: batch.error_file_id
    ? client.files.content(batch.error_file_id)
    : undefined,
});

openai.batch [task]
├── Chat Completion [llm]  input → output + tokens
├── Chat Completion [llm]  input → output + tokens
└── Chat Completion [llm]  input → error

[Luca Forstner (lforst)]

Abhijeet Prasad (@AbhiPrasad) refined the API a bit, see PR description.

[Abhijeet Prasad (AbhiPrasad)]

lets do it