Just going to talk about this with gaming since this is where I've seen this discussion the most. Honestly, I'd heard the primary issue with that whole thing was that you have to make a whole new model, rig said model, and then make animations for it. I'd say the latter probably takes the most time. Granted, I'm not exactly the best with this sort of thing, so I don't exactly know whether animations would be transferable between models. Furthermore, if the idea is to have expressive characters, then it's likely instead of transferring animations, one would opt to simply make a whole new set of them.
As for the issue of the issue between the way men/women walk, I suppose if you really went an analyzed it you might be able to find certain differences that come from biological differences at least for things like walking, but gestures and such are generally something that goes for an individual. Furthermore, there are people who would definitely go against the norm. Tying that back to animating, then the main issue there is the walk cycle I guess (that being said, i have actually animated a walk cycle before and, while it was kind of crappy, I wouldn't say it's hard thing to do).
So, in the end, I can kind of see where people are coming from when they lack a budget, team, or time in implementing this, but I don't see much of an excuse for bigger studios (well, unless we factor in the fact that they force their programmers into crunch time just to get some games out, but I think that's primarily blaming the publishers). Well, I guess there is an issue if you need to go back and change/rerecord dialogue, but that's more of a case by case thing.