One thing that's important to note about these stats outlined in this post is that pick rate =/= viability. Mythic diversity manifests mostly in what items are viable on a champion, and the items are different enough that certain gamestates call for certain items.
Using your own examples, Sylas (in the last 30 days) purchases everfrost in 88% of games. However, there are 3 other mythics with higher win % than it, one of which is Ludens with a 3% higher winrate. If players were more adaptable, they'd identify the situations where other mythics were more optimal than Everfrost and opt into those alternatives depending on gamestate.
The unfortunate problem with mythic diversity is that it's really tough for us to beat player psychology, which is to play what's comfortable or popular rather than critically think game-to-game. I don't blame players though, as up until this season there weren't meaningful decisions to be made so that skill hasn't been trained.