The short business answer is that it's currently not a metric that influences people's decision to install or uninstall an app. If it were one, we would've seen an engineering effort to reduce bloatware and duplications as seen in the post.
The trend towards huge installations is one of the many reasons I install as few apps as possible, especially in this class of "should have a fully functional web version" type social media things.
The damage is also cumulative: I have a perception that "apps" in general are likely to be huge piles of junk, so I'm reluctant to install any of them, regardless of their own merits. Even if LinkedIn were to spend the time improving their app it would suffer from the overall reputational damage to the entire ecosystem, and I still wouldn't even think to install it.