Neither Lucene nor Elasticsearch support a limit higher than Integer.MAX_VALUE, which is understandable.
A bit less understandable is the fact that Elasticsearch does not support an offset higher than Integer.MAX_VALUE, despite the fact they claim that a single index may hold more than that many documents due to the fact it's sharded: see the note near the bottom of https://www.elastic.co/guide/en/elasticsearch/reference/7.0/getting-started-concepts.html#getting-started-shards-and-replicas
Anyway, that's a very exotic use. Let's remove the methods accepting longs for now, we can always add them back later. Let's keep a long for the hit count, though: it's consistent with JPA, and I suspect Elasticsearch may still return a long in that case.