[alphazard]

Refactoring isn't an end on its own, and it shouldn't ever be considered a standalone project.

The easiest way to accomplish a real goal like fix a bug, or add a feature, may very well be to first refactor the code. And yes maybe you want to merge that in as its own commit because of the risk of conflicts over time. But just having the code look nice (to who exactly?) isn't valuable, and it encourages the addition of useless abstractions. It may even make later real-work harder. Never refactor outside the context of real-work.

The cartoon with the PM also gets at a ridiculous pattern: engineers negotiating when to do various parts of their job with non-technical people who have no idea how do any part of their job. The PM doesn't know what a refactor is, the EM probably doesn't either. It doesn't make the organization function any better to tell these people about something they don't understand, and then ask them when it should be done. Budget it as part of the estimate for real-work.