incremental reparsing in apollo-parser

[lrlna]

apollo-parser is using rowan for its syntax tree manipulation. rowan can be used to incrementally reparse various parts of the input without having to parse an entire schema or operation from scratch. We should explore various ways we can have apollo-parser do the same.

A basic use case: let's say we have a fairly large schema with 500 definitions and a user decides to add 1 more definition to make it 501. How do we only parse that one extra definition?