Freeciv

The current behavior is bad, but this is a bit harder to fix than expected (I've looked at it every now and then, but have no real progress). This might be something that we just need to accept also in 3.1, and not block S3_1-d3f much longer by this. If no objections (or better; a fix) soon, I'll postpone this.

With most former S3_1 d3f blockers that we've postponed to 3.2, we've had clear plan how to implement them before S3_2-d3f (and in most cases we've actually implemented them to master already). This one is very frustrating in that I don't know how it would get fixed before 3.2 either, so postponing is sort of not resolving anything. Would be much happier to ignore this for now (while concentrating to S3_1-d3f) if I knew that it will not then turn up exactly same way when trying to do S3_2 d3f.