fix(ai-help): add short and extended explanatory guidance

[LeoMcA]

Summary

#9214
#9208

Problem

We don't explain that outputs from AI Help are not guaranteed to be accurate, and that the links to the docs used to generate the prompt exist to validate the response.

Solution

Add a short explanation of this, which expands into a modal containing a longer one.

---

Screenshots

Before

After

---

How did you test this change?

yarn dev, tested across different screen sizes - did not test with responses in the chat, I guess our testing creds changed, so could potentially mess with autoscroll being a bit off, but I reckon this is ok for now

[fiji-flo]

Thanks for doing this on a Sunday. Looking good 👍

[MrPetovan]

The problem of this features wasn't the lack of explanation about the potential inaccuracy, the problem is the inaccuracy itself.

You're operating a reference documentation, and you rolled out a feature encouraging to "cross-verify" the information this feature produces with your own information? This doesn't make any sense.

[oscherler]

I’m glad it’s about a web site and not a zoo. Imagine a lion escaping its cage, and they just put up a sign at the entrance saying “Caution, loose lion on the premises.” instead of closing the zoo until the lion is captured.

Can you please update the PR title to remove the word “fix,” and use something more accurate like “temporary mitigation?”