You can't optimize your ads for conversions until you've already had a minimum number of conversions (I think 30 per month?). Google can still run up hundreds of dollars of paid clicks that will never result in conversions before then (and mobile apps for kids are one way to do that).
As for desktop vs tablet, some people might be advertising a tablet-only app or a desktop-only app. In that case it's best to advertise on the device people can use it on, to streamline the download/purchase.
That's true. Still doesn't mean you should optimize for CTR when you're really interested in conversions. But I do see your point that in some cases - especially when starting a new campaign on a tight budget, it would be useful to manually target the campaign with more precision (including tablets vs desktop control and more obvious control for apps). Ideally, Google would allow people to start optimizing for conversions from the beginning, but I guess their algorithms have to start with something (clicks being the most abundant form of response data).
As for the desktop-only or tablet-only app, I'd still rather let that work itself out with conversion optimization (given that download conversions can be tracked effectively) rather than assuming my conversions are going to come through a certain channel. It comes down to the idea of making marketing decisions based on real data rather than what we're "sure" is best.
It just sounds like many people don't bother setting up conversion optimization out of ignorance - they think Google is somehow doing this app/game placement just to try and steal people's money, when really it's just a natural consequence of the CTR optimization. In the case of Aha!, it appears that most of the "wasted" money was spent before even realizing what was going on - and once they did figure it out, they apparently found the way exclude mobile apps without too much trouble (though I'd agree Google should make it easier).
Oh definitely, if you have the option to optimize for conversions, you should definitely try it. From memory that was a key factor in patio11 finding success with Bingo Card Creator.
That said, if your conversion rate is the typical "1% of trial users buy / x% of visitors download", it can take a long time to optimize. And if you try to use trial signups / downloads as a proxy conversion for sales, you can end up optimizing for the wrong thing (eg optimizing for freeloaders / kick the tyres, not buyers).
Curious to hear if Aha!'s mobile app exclusion method actually works - I tried a similar exclusion and it didn't work for me, I still play mobile app whack-a-mole. Google called me when I stopped my ads & told me that entering a -300% bid for mobile was a hack from engineering that would stop showing all ads on tablets, but the interface wouldn't accept anything more than -100% (which I already had).
As for desktop vs tablet, some people might be advertising a tablet-only app or a desktop-only app. In that case it's best to advertise on the device people can use it on, to streamline the download/purchase.