Disconnect between semantics of 'updated_at' and actual updates

Summary

We're tracking stale issues, but we're finding that updated_at is bumped when an issue is referenced, but it's due to a system note created. This doesn't feel like a real update but the issue is no longer stale.

There's a bit of disconnect between what humans would consider an update and what GitLab's database considers an update.

Steps to reproduce

Create two new issues, A and B. Reference #B from A, and the updated_at is bumped due to the system note even though that issue wasn't 'updated'.

What is the current bug behavior?

The issue 'updated_at' might be used for tracking stale issues, and this is polluted by system notes unrelated to actual updates, making stale issue tracking tricky.

What is the expected correct behavior?

I'm not sure what the answer is, but perhaps a different column, or a way of filtering out system notes?