Use better heuristics for when the media should be sent as a document

[Veetaha]

The simplest heuristic right now is detecting some fixed large resolution with comic tag (or maybe text tag?). Because it's very likely to be unreadable if sent as a photo.

Anyway, this isn't yet clear. This needs more research on the design of the heuristics.