Logging consistency and safety

When writing/adjusting code, treat logging as part of error handling:

• Use the right level: log request- or operation-failure conditions at error/exception (not warning).
• In except blocks, prefer get_logger().exception(...) (or error(...)) and include identifying context (IDs, URLs, comment_id, PR/issue number) so logs are traceable.
• Use structured/context fields (e.g., artifact={...}) rather than passing arbitrary objects (like dicts) as separate arguments to the logger—this can cause formatting/type issues.
• Avoid print statements for debug; use get_logger().debug(...) with the same context.
• Don’t comment out logging used for exception diagnostics—either log it or remove the dead code.

Example pattern:

try:
    ...
except Exception as e:
    get_logger().exception(
        "Failed to publish inline code comments fallback",
        artifact={
            "comment_id": comment_id,
            "error": repr(e),
            "pr_number": pr_number,
        },
    )
    raise

For non-exception unexpected formats:

get_logger().warning(
    "Unexpected response format",
    artifact={"response": str(response_tuple)},
)