Investigate decorating the lacinia schema

[oliyh]

Imagine if a resolver B in the schema could also be given a bucket id, and all enqueues from that resolver would go into that bucket. If the parent resolver A could look ahead and see that B will be called as a child (one or more times) it should be possible to size the bucket for B based on the result of A * fields resolved by B.

I don't know if the context passed to a resolver contains enough information about the children to be able to get the bucket for B, or find how many child fields might use it.