Add L1 client metrics

[jbearer]

Adds visibility into the L1 client. In retrospect these are the metrics we would have needed to automate monitoring for the issue we observed recently with stuck L1 clients on decaf. We can add alerts on consensus_l1_head increasing to catch issues like this faster in the future.

This PR:

Adds the following metrics to give visibility into the L1 client:

• consensus_l1_head: the block number that the node thinks is the current L1 head
• consensus_l1_finalized: the block number that the node thinks is the most recently finalized L1 block
• consensus_l1_ws_reconnects: the number of times the node has had to reestablish a WebSockets connection
• consensus_l1_stream_reconnects: the number of times the node has had to reestablish a blocks filter or subscription

Logs L1 RPC errors at RPC client layer. This ensures all errors get logged straight from the RPC, even if they are ultimately swallowed at a higher level (as in the Ethers FilterWatcher stream). This will give us much better insight into future problems with the L1.