Whups! I had a mistake when I originally submitted - and thought it was a failed experiment... Fixed that up and now the slotmap is _much_ faster :) Current b

Whups! I had a mistake when I originally submitted - and thought it was a failed experiment... Fixed that up and now the slotmap is _much_ faster :) Current benchmark has it beating a native-map based approach by 2x in a mock real-world scenario