feat: Return logits_prob in `chat_completion`

[hiro-v]

Problem

• Hi, I'm using https://github.com/EleutherAI/lm-evaluation-harness/tree/main to benchmark for MMLU and TruthfulQA. However currently cortex does not return logits_prob

Success Criteria
A clear and concise description of what you want to happen.

Additional context

• https://github.com/ggerganov/llama.cpp/issues/6285
• https://github.com/ggerganov/llama.cpp/issues/6423

[dan-menlo]

@hiro-v Great idea - merged into #1151 for Sprint 21