Feature/9002 optimise reranking algorithm

[shaunahu]

1. remove dataset_group from should clouse to sort option. So a matched dataset_group not contribute to the ES relevance score, but used for priority sorting. This aims to reduce the noises of current method, which gives matched dataset_group query a very high boost, and resulting in a matched organisation record even though have low relevance, it is still in the search result list.

2. update the explain simplified response structure, to include sort options. So that easy to understand which field prioritised the record.

3. use a shared createKeywordShouldClauses for both explain and search to build ES query

[NekoLyn]

LGTM