Batch processing in cloud functions sucks

[OR13]

Cloud functions need to load all state needed to operate.

the batch handling that sidetree normally uses, does not map well to cloud functions, especially with our current database integrations and syncing between couchdb and the in memory rxdb instance.

We don't want to start each cloud function execution by syncing a whole data base (we aren't doing this, but we need a workaround for improving perf and avoiding infura calls for cloud functions).

We need to implement a direct to couchdb querying interface for element, and provide a better seperation of concerns between offline first mobile pwa nodes, just in time cloud function full nodes, and always on full node servers (the traditional case).