As long Canonical maintain the backends, it's probably no problem for them to get them upstream in gtk and qt. Rim has as I understand it the blackberry backend merged upstream in qt.
I think there is a decent chance of it getting into Qt if it is written well enough. But Qt 5 is designed to be highly modular -- it is easy to add or remove backends, so other distros can disable building the Mir backend, and Qt can get rid of it easily if canonical stops supporting it. Further, Canonical will be using Qt a lot. So there is very little downside, and it will likely encourage Canonical to make more use of Qt. But that is probably the extent of it, there will likely be little if any additional work done to support it in KDE, for example.
I don't know as much about gtk, but I am under the impression that it is not as modular. This would mean that adding a new backend is a major job that affects a lot of other things and will make removing it much more difficult. If that is accurate, and considering the recent bad blood over Unity vs. Gnome Shell, I think the chance of the gtk backend being upstreamed is much smaller.
10-100 people isn't a sufficient sample size for a population of 10-100 million. Also, sample size isn't the only problem with personal experience. There's also the problem of confirmation bias.
100 people is sufficient to sample a population of any size (be it a thousand or several billions). You'll have trust intervals on any indicator you get out of the sample (but you'd have them regardless of the size of the sample, and the size would almost not depend on the size of the population). But those would be largely small enough for testing any hypothesis that begins with "most of", for example.
Again, the issue is "is the sample representative of the full population?", and the answer is no. Confirmation bias is only a sub part of it. If you only take your sample in your country, you're still not representative, even if you don't limit yourself to people that think like you.
I'm not saying he's right, or his experience relevant. I'm specifically saying that your claims about sampling are imprecise and/or wrong.