changes to keep the ws connection alive with a ping

[fbielejec]

Summary

This PR addresses a couple separate problems:

1. Infura drops idle ws connections after approximately 60 seconds, unfortunately quietly for the client, never triggering on-disconnect

2. Infura uses a cache of events, but presumably only for some of the web3 RPC calls. This manifested because when processing events we use eth_getblockbynumber to get the event timestamp.
   This would throw a fatal exception (only when processing latest events), because even though they arrived the block was presumably not yet cached by infura.

3. SqliteError: NOT NULL constraint failed: param_changes.param_change_SLASH_original_value #664

Ad 1) is addressed by altering the server-web3-library and adding a continuous keep-alive poll. Should it ever encounter an exception, the system goes into an offline mode and starts polling for the connection to go back online. For monitoring purposes I started logging the number of resyncs due to websocket connection dropping. We also detect the on-disconnect event but as shown before it's not to be trusted 100%.

Ad 2) I added a wait-for-block function in server-smart-contracts library, which polls until the block is availiable.

Ad 3) changed the model (reason field can be nil)