Make OpenAI Responses token ceilings templated and surface truncation

[Lee-W]

Dags need a per-run token ceiling that can vary by environment, but max_output_tokens and max_tool_calls were only reachable through the non-templated response_kwargs dict. Promote both to first-class, templated operator arguments, coerce the rendered string to a positive int before the request goes out, and reject an invalid or non-positive value instead of silently running with no ceiling. Also reject the same key being set both as an argument and inside response_kwargs, so one silently wins.

OpenAI's Responses API has no monetary cost limit to expose, so this is a token ceiling only -- for a cost cap, use
apache-airflow-providers-common-ai instead. Hitting max_output_tokens does not fail the request; the response comes back incomplete with truncated output text, which the previous 'may be empty' warning did not describe. Surface
incomplete_details.reason in the log message so operators can tell a truncation apart from other non-completed statuses.

---

Was generative AI tooling used to co-author this PR?

• Yes (please specify the tool below)

---

• Read the Pull Request Guidelines for more information. Note: commit author/co-author name and email in commits become permanently public when merged.
• For fundamental code changes, an Airflow Improvement Proposal (AIP) is needed.
• When adding dependency, check compliance with the ASF 3rd Party License Policy.
• For significant user-facing changes create newsfragment: {pr_number}.significant.rst, in airflow-core/newsfragments. You can add this file in a follow-up commit after the PR is created so you know the PR number.