Detect unsynchronized state of RPC node to use the fallback

[akolotov]

The currently the oracle is not able to identify that requesting a data from an unsynchronized RPC endpoint that is why it could lead to some undesirable issues. For example, the sender worker will try to call eth_getTransactionsCount to get the current nonce to prepare a transaction but the value will be lower than the actual nonce stored in the blockchain that is why the transaction disappears after sending (an error that the nonce is lower than expected will not be raised since the transaction will be send to unsynchronized enpoint).

Consider to implement to mechanism to discover that the RPC endpoint is unsynchronized, report about it in the logs, and use a fallback RPC in the operations. Another requirement to this feature is to make as less RPCs requests as possible identify the unsynchronized state.