Introduces a Renderer type and a DocumentRenderer implementation which generalizes the the rendering of a Document so that it can work with both query and schema documents (the inverted parser).

Running the benchmarks:

// Without document caching
[info] Benchmark                             Mode  Cnt       Score      Error  Units
[info] GraphQLBenchmarks.renderCaliban      thrpt    5  103325.148 ± 5125.692  ops/s
[info] GraphQLBenchmarks.renderCalibanFast  thrpt    5  235363.260 ± 1367.168  ops/s

// With document caching
[info] Benchmark                             Mode  Cnt       Score      Error  Units
[info] GraphQLBenchmarks.renderCaliban      thrpt    5  106606.129 ±  511.746  ops/s
[info] GraphQLBenchmarks.renderCalibanFast  thrpt    5  842012.538 ± 1521.568  ops/s

The above benchmarks show a 2x improvement over the base rendering (calling GraphQL#render) when using .toDocument, however, if we cache the document (something I added on this PR) we see an 8x improvement on rendering.

This should help with use cases that require the document to be re-serialized (e.g. federation, stitching, etc.)

TODO:

• Finish Compact parsing
• Migrate remaining "rendering" components
• Add more tests