its primary purpose is to lay the foundation for “modular hivemind” (an application framework for 2nd layer applications)
That is a killer - 2nd layer has lot of potential - literally app can store anything in JSON and process as needed - once it's ready, we should plan for publishing some ready to run sample codes in various languages, that can help developers to set up a second layer app in few hours. I hope the data that's going to be stored in second layer, is not going to hurt the performance of the chain, right ? Also is there a limitation on the size of data that can be stored in second layer ?
A full-scale example app is planned as a deliverable for modular hivemind. Most likely the example will be a wallet app.
I don't expect 2nd layer data to have any major impact on the chain. From a bandwidth perspective, RC limiting will prevent traffic congestion. Most of the computation takes place at the 2nd layer in this model, and the design is such that not every modular hivemind node will need to run every 2nd layer app (for example, in the case of the light wallet, users could just run their own local copy even). This means that each modular hivemind installation gets to determine what data it wants to keep, and it only needs to keep the data for the app or apps it wants to support.
There's no overall limit on 2nd layer data other than that imposed by RC limits, but there is a limit on the size of a single 2nd layer transaction (currently it's 8K per transaction, IIRC).