fix: Fireworks chat completion broken due to telemetry

[slekkala1]

What does this PR do?

Fix fireworks chat completion broken due to telemetry expecting response.usage
Closes #3391

Test Plan

1. uv run --with llama-stack llama stack build --distro starter --image-type venv --run
   Try

curl -X POST http://0.0.0.0:8321/v1/openai/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
      "model": "fireworks/accounts/fireworks/models/llama-v3p1-8b-instruct",
      "messages": [{"role": "user", "content": "Hello!"}]
    }'

{"id":"chatcmpl-ee922a08-0df0-4974-b0d3-b322113e8bc0","choices":[{"message":{"role":"assistant","content":"Hello! How can I assist you today?","name":null,"tool_calls":null},"finish_reason":"stop","index":0,"logprobs":null}],"object":"chat.completion","created":1757456375,"model":"fireworks/accounts/fireworks/models/llama-v3p1-8b-instruct"}%   

Without fix fails as mentioned in #3391

[franciscojavierarceo]

lgtm

[mattf]

$ curl https://api.fireworks.ai/inference/v1/chat/completions -s \
-H "Content-Type: application/json" \
-H "Authorization: Bearer ... \
-d '{
  "model": "accounts/fireworks/models/kimi-k2-instruct-0905",                  
  "messages": [{
      "role": "user",
      "content": "Explain the importance of fast language models"
  }]
}' | jq
{
  "id": "c07ea231-a59d-4828-a169-a8e4243f907f",
  "object": "chat.completion",
  "created": 1757520350,
  "model": "accounts/fireworks/models/kimi-k2-instruct-0905",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 23,
    "total_tokens": 727,
    "completion_tokens": 704
  }
}

the fireworks endpoint returns usage information. before making a project wide change, especially one that may result in quietly inconsistent results, a fix must be attempted for the fireworks provider.

@raghotham @franciscojavierarceo i recommend reverting and proceeding w/ a fix in the fireworks provider.

[franciscojavierarceo]

@mattf I've created a revert PR here: #3402

[slekkala1]

$ curl https://api.fireworks.ai/inference/v1/chat/completions -s \
-H "Content-Type: application/json" \
-H "Authorization: Bearer ... \
-d '{
  "model": "accounts/fireworks/models/kimi-k2-instruct-0905",                  
  "messages": [{
      "role": "user",
      "content": "Explain the importance of fast language models"
  }]
}' | jq
{
  "id": "c07ea231-a59d-4828-a169-a8e4243f907f",
  "object": "chat.completion",
  "created": 1757520350,
  "model": "accounts/fireworks/models/kimi-k2-instruct-0905",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 23,
    "total_tokens": 727,
    "completion_tokens": 704
  }
}

the fireworks endpoint returns usage information. before making a project wide change, especially one that may result in quietly inconsistent results, a fix must be attempted for the fireworks provider.

@raghotham @franciscojavierarceo i recommend reverting and proceeding w/ a fix in the fireworks provider.

@mattf Thanks for the suggestion!

Well, it didnt return usage in my test with fireworks provider. (May be I miss something in provider impl, I can have a second look)

Would it ok for the api to be broken because the telemetry depends on response to have certain fields?

[mattf]

$ curl https://api.fireworks.ai/inference/v1/chat/completions -s
...
the fireworks endpoint returns usage information. before making a project wide change, especially one that may result in quietly inconsistent results, a fix must be attempted for the fireworks provider.
@raghotham @franciscojavierarceo i recommend reverting and proceeding w/ a fix in the fireworks provider.

@mattf Thanks for the suggestion!

Well, it didnt return usage in my test with fireworks provider. (May be I miss something in provider impl, I can have a second look)

Would it ok for the api to be broken because the telemetry depends on response to have certain fields?

api.fireworks.ai returns the info, if it doesn't get propagated then check the fireworks provider.