Proposal of restructuring code (RFC) #570

As the project grows, in order to better manage the code, expectations and dependencies, I hope we can re-organize the repository by creating multiple modules. With that, we no longer have to put all ideas into one single lib, trying to make it both a JDBC driver and Java client for ClickHouse. Instead, we'll follow JDBC specification in clickhouse-jdbc module for standard update/delete operation, synchronous execution, and escape syntax like {d '2020-02-17'}(which is same as toDate('2021-02-17') by the way) etc. On the other hand, we'll stick with the "clickhouse-way" when implementing clickhouse-client module, so that it behaves like any other clickhouse clients.

• optionally new package name(and maven groupId): ru.yandex.clickhouse -> tech.clickhouse
• new directories(and maybe a separate repo like clickhouse-java later)

  |
  |-- clickhouse-client
  |-- clickhouse-http-client
  |-- clickhouse-grpc-client
  |-- clickhouse-native-client
  |-- clickhouse-jdbc
  |-- clickhouse-jdbc-ext (tentative)
  |-- clickhouse-gql (tentative)
  |-- clickhouse-r2dbc (tentative)
  |...
  

• modules and API layers
  
• how to approach?
  I hope we can implement the new structure starting from 0.3.0, but it's primarily about moving code from src to clickhouse-jdbc/src and pile up code into other modules. This ensures we can produce binary compatible package in 0.3 series. And then, starting from 0.4.0, I hope we can optionally change package name and refactor clickhouse-jdbc module by leveraging code from clickhouse-client module etc., which will introduce different binary packages.

Please feel free to comment and I'll start to create PR if no concern.