Fix bug when delayed index streams are added to the query plan multiple times

[apmoriarty]

Fixes https://github.com/NationalSecurityAgency/datawave/issues/1643

• queries should no longer duplicate subtrees when delayed terms exist
• document range query plans now have delayed terms propagated correctly
• fix rare case that would skip shards for queries like A && (B or delayed(C)) by adding an IndexStreamComparator to ensure that VARIABLE children are advanced after a context has been found
• RangeStream correctly handles a top level union with a VARIALBE context
• Intersection and Union report more accurate stream contexts
• IndexStreams support nexting with context
• ConcurrentScannerInitializer builds index streams with correct context