Nobody is saying teams can't be lucky, if you are then great, but you can't keep getting lucky. And that's not how the stat is used. xG is a long-term measure, all teams eventually trend to their xG over a high enough volume of games. With a long time-horizon luck becomes less and less of a factor because your sample size grows. That's just how statistics works. You can only get lucky for so long. Like with United - anybody with eyes can see they have a big problem with performances and are papering over the cracks with lucky goals and narrow wins. Their xG demonstrates this perfectly, no other stat has that descriptive power.
That's one of the more obvious cases though, I worked with a League 2 team who were top after 5 games and thought they were the dog's bollocks, they refused to change approach even though their xG said they hadn't justified their wins - they'd gotten the rub of the green with ref decisions. Because they stubbornly stuck to what got them those wins, they ended up avoiding relegation by only 2 points. The law of averages caught up with them. This is also how casinos ensure they have a house edge. xG is about ensuring your team has the house edge. City always always have the house edge, that's fundamental to how we play.
You can call it bollocks all you want, I've worked as a data analyst at multiple football clubs - some league and some non-league and I'm telling you they don't think it's bollocks. Your question "how can you have 0.78 of a goal?" is total ignorance of what it actually represents. 0.78 xG means that if you had that shot (or combination of shots) on goal 1,000 times then you would score on average about 780 times. We're not talking fucking quantum physics here.