We want the Moderator Tools team to think about this first

I can see the benefit of this. We briefly considered something like this for Automoderator - such as adding a flag to the recentchanges table to highlight that Automoderator hasn't reviewed the edit yet, so that patrolling tools like Huggle could avoid displaying it, which may be a waste of patroller time if Automoderator then reverted it. Ultimately Automoderator's time to revert is on the order of seconds, so the potential impact seemed pretty low.

I think this would probably be best at the patrolling-feature level, not the page level, i.e. the Android patrolling feature could avoid displaying the same edit to multiple users in a short period of time. It would be hard to collate all the different ways an edit might be reviewed (Anti-vandalism bots, external tools like Huggle, Recent Changes, Android patrolling ...) into one "someone's looking at this" data source.

The only downside I see is that it may be hard to tell the difference between someone who opened an edit in the patrolling feature and put their phone down and someone who opened the edit and then went to search the internet to verify the change. If we can't tell that someone is actually actively reviewing an edit we risk letting bad edits sit around not actually being reviewed by anyone. Perhaps prioritising the edits you show to a user in the patrolling tool based on how many other users they've recently been shown to could be one approach to minimise this situation from occurring.