Add concurrent access documentation to BufferAggregator.relocate()

As far as I can tell, currently it should say `relocate() is used only in GroupBy implementation in a single-threaded environment; it cannot be called concurrently with get() or aggregate()`, however, @himanshug's growable proposal (https://github.com/apache/incubator-druid/issues/6743#issuecomment-509415145) assumes this method to be called during indexing as well, so it might make sense to add future-proof propositions.

/cc @jihoonson

Locate BufferAggregator.relocate() and read its current documentation alongside get() and aggregate(). Review the linked growable proposal in issue 6743 for the indexing context; done means the method’s concurrent-access constraints and any agreed future-proof wording are documented clearly.

Written by the indexing model from the issue text.