Add an explicit AI policy to our documentation

As noted in #13825, we're starting to get AI-generated contributions. In order to ensure our limited maintainer bandwidth isn't overwhelmed, and to protect us from any ownership or licensing concerns over submitted code, I suggest that we explicitly adopt and document an AI policy. My strong preference is for something based on the `attrs` policy - https://github.com/python-attrs/attrs/blob/main/.github/AI_POLICY.md. In my view, this takes the correct stance, focusing on ensuring that every PR has a human contributor associated, and that contributor takes full resaponsibility for the technical and legal/ownership aspects of their contribution.

There are always going to be judgement calls involved - after all, LLMs *by design* produce plausible looking content - so the policy will of necessity be subject to interpretation. By choosing a policy that assumes good faith on the part of submitters, while still being clear that we *will* impose penalties like rejection or banning when necessary, I think we strike a reasonable balance.

I have created #13826 as a starting point for discussion.

[]