Adding support for JinaAI late chunking

[dan-rubinstein]

TODO

Questions:

1. In Elasticsearch we limit chunking to at most 512 chunks. When running through the EmbeddingRequestChunker we provide a maxNumberOfChunksPerBatch that the third-party service can handle. For late chunking, this value has to be at least 512 as we must pass all of the chunks for a given document into a single call downstream. Is this an assumption we can make?

[elasticsearchmachine]

Hi @dan-rubinstein, I've created a changelog YAML for you.