fix runtime exception for deep async queries

[jbellenger]

Problem
Resolving large queries with more than 16 async levels throws this error:

java.lang.IllegalArgumentException: unknown level 16
	at graphql.execution.instrumentation.dataloader.LevelMap.get(LevelMap.java:35)
	at graphql.execution.instrumentation.dataloader.FieldLevelTrackingApproach$CallStack.allStrategyCallsHappened(FieldLevelTrackingApproach.java:65)
	at graphql.execution.instrumentation.dataloader.FieldLevelTrackingApproach.levelReady(FieldLevelTrackingApproach.java:238)
	at graphql.execution.instrumentation.dataloader.FieldLevelTrackingApproach.dispatchIfNeeded(FieldLevelTrackingApproach.java:223)
	at graphql.execution.instrumentation.dataloader.FieldLevelTrackingApproach.handleOnFieldValuesInfo(FieldLevelTrackingApproach.java:177)
	at graphql.execution.instrumentation.dataloader.FieldLevelTrackingApproach.access$000(FieldLevelTrackingApproach.java:26)
	at graphql.execution.instrumentation.dataloader.FieldLevelTrackingApproach$1.onFieldValuesInfo(FieldLevelTrackingApproach.java:147)
	at graphql.execution.AsyncExecutionStrategy.lambda$execute$1(AsyncExecutionStrategy.java:73)

The issue appears to be that inside FieldLevelTrackingApproach.levelReady, we're checking the allFetchesHappened LevelMap when it may not have been expanded to the specified level. This is not an issue for small queries that can fit under the default size of 16 levels, but throws an exception for more nested async queries. This issue effects all "happened"-ish LevelMaps in FieldLevelTrackingApproach

Solution
In LevelMap.get automatically resize if needed. This allows deep queries to match the behavior of shallow queries which do not need to resize their LevelMaps.

[bbakerman]

Cool - some more lines ;)

[bbakerman]

Thank you for this