Lucene: Avoid GPU OOMs and also Java-heap OOMs

[nvzm123]

This fixes two issues:

1. We were eagerly loading the full dataset into GPU instead of letting CAGRA manage streaming it to GPU.

2. We were passing full datasets to createMultiLayerHnswGraph which crashed my java heap space when benchmarking 10M 1536d (~60GB) dataset; now we can stream manageable subsets instead.

PR: NVIDIA/cuvs-lucene#141