Silence tracebacks for rate limit and connection errors in get_response

[hemanth-asirvatham]

Motivation

• Reduce noisy full-traceback logging for recoverable RateLimitError and APIConnectionError so the first-occurrence notices handle user-visible output and subsequent occurrences are silenced and tracked in periodic status updates.

Description

• In src/gabriel/utils/openai_utils.py updated both API-call branches in get_response to catch RateLimitError and APIConnectionError and log them at DEBUG (no exc_info) before re-raising, instead of logging full error tracebacks at ERROR.

Testing

• No automated tests were run for this change.

---

Codex Task

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}