Refactor engine management logic into a separate library

Right now, Skipchunk has facades for index management, and generic querying, in Solr and Elasticsearch.  This is used by Skipchunk itself, but also now by Hello-NLP.  It feels odd that this lives in Skipchunk and the functionality may have goals beyond skipchunk's.  So engine/index management/admin would be better maintained and grown by existing in a separate library.