Explore possible chunking optimisations

This ticket is set up to explore any optimisations possible when creating chunks when indexing or using them at query time.

The default size of a chunk in zapx is 1024 items, since that's believed to be the optimal number of items needed to be "walked" without skipping(Ref: https://github.com/blevesearch/zapx/blob/master/chunk.go#L53-L61). .

The first experiment is to vary the chunk size and see its impact on specific query types like conjunction, disjunction and geo at both query and indexing time(More details in comments).

Will continue to update this as I explore more ideas to improve our chunking and related capabilities.