Sort:  

dApps can now design their custom_json op related features and have access to data easily via endpoints, which I offer help to design. For example, if Splinterlands was a new game coming onboard, they could opt to use Plug & Play (or a customized version of it) to handle all the game data generated by users.

It makes it easier for developers to make apps, without handling much blockchain code.

So for example I want to write a new dapp, that would persist a post to the main blockchain and then some custom json - so I would have to do two operations ? one for the main one and another for custom json ?

Yes, you will have to do two operations.

How Plug & Play helps is on accessing the aggregate custom_json data that your dApp deals with, using PostgreSQL tables/views and customized API endpoints.

So you won't have to stream blocks to get those custom_json ops or run your own server if you don't want to.

Without Plug & Play, you will have to stream whole blocks from the first block that your app started, using a library like beem, to populate your own tables and spin up your own server to avail API endpoints.