storage refactor

[chunyang-wen]

Storage should provide at least two levels of abstraction:

• The storage level abstracts the details of the corresponding storages such as rockdb or other database
• In the middle, we construct graph from database
• Graph engine level converts data compatible with the specific query engine, such as convert to LynxInteger

[merlintang]

good point

[LianxinGao]

Currently, in Storage level, we have a trait org.grapheco.tudb.store.storage.KeyValueDB to abstract all the key-value DB ( it can be redesign～)