v2 LookupSet

[austinabell]

Unlike current LookupSet, API should match HashSet as much as possible (although most functionality won't be possible until UnorderedSet is implemented because keys won't be tracked).

Similar to #487 except instead of writing serialized values, it'll be writing empty bytes and checking just if storage exists for the given keys. I think keeping the hashing functionality makes sense, to keep key length consistent