feat(openai): add logprobs support for chat completions (request encoding, response decoding, tests)

[jat10]

Description

Adds openai_logprobs and openai_top_logprobs provider options to the OpenAI schema. When enabled, logprobs are encoded in the request body and extracted from choices[0].logprobs.content in the shared decode_response_body_openai_format, making them available as response.provider_meta[:logprobs] for all OpenAI Chat Completions responses.

Type of Change

• Bug fix (non-breaking change fixing an issue)
• New feature (non-breaking change adding functionality)
• Breaking change (fix or feature causing existing functionality to change)
• Documentation update

Breaking Changes

NA

Testing

• Tests pass (mix test)
• Quality checks pass (mix quality)

Checklist

• My code follows the project's style guidelines
• I have updated the documentation accordingly
• I have added tests that prove my fix/feature works
• All new and existing tests pass
• My commits follow conventional commit format
• I have NOT edited CHANGELOG.md (it is auto-generated by git_ops)

Related Issues

[mikehostetler]

Thanks for adding the logprobs support and the solid test coverage around request encoding, non-stream decoding, and streaming accumulation. I marked this needs_work for one small validation gap: openai_top_logprobs is documented as 0-20, matching the OpenAI API, but the schema currently accepts any non-negative integer, so values like 21 pass local option processing and get encoded into the request. Please tighten the schema or add validation for the 0-20 range. Nice addition overall; this should be straightforward to finish.

[jat10]

@mikehostetler Fixed, changed openai_top_logprobs type from :non_neg_integer to {:in, 0..20} so NimbleOptions rejects out-of-range values before they ever reach encoding. Added boundary tests for 0, 20 (valid), and -1, 21 (rejected).

Once this merges, I'll open a PR on jido_ai that bumps the req_llm dep and wires up logprobs