@CKoerner_WMF, if you search for a place in Google, e.g. New York, you see this result (image below). Clicking (or double clicking) the map in the upper right corner would show google maps full screen. What we have is somewhat similar - we have an article with a link (snapshot) to the map. Bing.com has exactly the same behavior - you will have to scroll down a bit to see the map in the results, but it is also an image, not an interactive map. So does yandex.com. In other words, the large sites follow this model due to the general reading pattern - you view information, you get interested in something which you click - and that makes it full screen.
We do not have to follow the same model, and come up with some sort of an "in-place activation" (clicking makes it interactive), but it might not be very intuitive for the users, especially if the map size (controllable by the editors) is small. We cannot make it interactive from the start because it will force every user to download lots of extra code instead of getting just one image when they don't care about interacting with the map (most of the time). We could allow interactive map from the start only on the smallest of the wikis - like WikiVoyage, and even there it might make sense to show a non-interactive image first - but this is more difficult because it would break what users are already accustomed with.

@CKoerner_WMF I have been thinking some more about this - I think that while we keep Wikivoyage mapframes interactive, we should make double click into a zoom in. The rest of our wikis that use static snapshot service should only allow a single click that becomes a full screen. I am not too happy about this difference between WV and the rest of the wikis, but I suspect that the vast majority of our users have never been to WV, so would not notice this difference. At the same time, this would be consistent with how most web sites (especially search engines) show information + image of the map, with a click required to activate the map. What do you think?