My comment is an observation about how this gets tried every few years in major libraries and is usually reverted. I Agree, there are use cases where these are better. But the pattern tends to be to revert to simpler data structures.