Manage long-running queries

This is probably a discussion piece in the first instance.

It is common for organizations who run CiviCRM to run a job on the server that kills queries if they run for more than 10 minutes or so. There is some demand that we manage this better. I'm going to paste in the feedback I have so far:

"We run a cron job to kill long running mysql queries. It would be good if CiviCRM could enforce such limits itself, rather than everyone implementing their own watchdogs."

Different org

"There's a "slow query killer" script running on the mysql host. I guess what we do is,

manually notify person whose query was killed, and scold them perhaps if it's a behavioral thing.
Occasionally scan the slow query by eye and investigate egregious things. Having the username helps us figure out what happened."

(I'm uploading the patch that makes the second of these possible.)

It might also be valuable to the project if there was some way to get error reports on this sort of thing.