Extend SpamCheck API?

[matrixbot]

This issue has been migrated from #10311.

---

Recent experience suggests that we would benefit from having some more details about messages (and invite requests, etc.) to determine whether a request is spam.

• We have the IP of the user during registrations. But if we only detected that the IP is used for spam after registration, we can't block them through the antispam. Ideally, the antispam should have access to both registration IP and current IP, assuming that the user is on the current antispam.
• Some antispams may wish to apply policies such as "can't invite if you have just registered", etc. So, having access to information on when the user has registered would be useful.
• Similarly, how many messages they have already sent that weren't picked as spam.