Synchronizer error handling

[chancehudson]

When there is a connection when retrieving blocks it doesn't restart. This forces the user to restart the browser window to get the synchronizer to continue downloading blocks.

[chancehudson]

The synchronizer makes lots of calls to a web3 node that can potentially error, like here and here. These calls should be protected by some try/catch/retry logic to make sure that synchronization continues if there is a brief disruption to the connection.

There's also lots of repeated logic in the synchronizer around loading events from the past, this can probably be refactored at the same time.