I don't think we should do this for suggestions, because it makes them significantly less lightweight. Since you can choose whether or not to engage with suggestions in the first place, and they aren't blocking like pre-save checks are, a survey being present at all will reduce the amount of "dismiss" signals we get from users by making them more likely to just ignore suggestions and move on to publishing. This will give us less overall useful data about suggestion-effectiveness.

Per offline discussion, before prioritizing work on implementing a survey of this sort, we need to align on whether one is warranted.

For now, I think T401739 is a sufficient way to gather feedback. We can revisit the priority of implementing a mechanism of this sort in the context of default-on deployments (T414853).