Fast block/blob archive node via ERA files

[dapplion]

If a user wants block archive data (all blocks from genesis to head) or blob archive data (all blobs from deneb to head) the network does not naturally serve that data as nodes are not required to keep that data. Discovering other archive nodes to sync from is not guaranteed and sync times are very slow.

ERA files (spec: https://github.com/eth-clients/e2store-format-specs) are a standardized way to distribute block and blob history that we can leverage to bootstrap archive nodes. The process is:

• Download ERA files (from URL, IPFS, torrent)
• Load and verify them
• Reconstruct intermediate states to get a fully synced archive node

However, turns out that writing 800 GB into LevelDB is not a good idea. The node spends (90% of the load routine compacting) and there's a huge write amplification. So this is a good opportunity to switch Lighthouse cold DB backend to static files (like Reth, Geth, and others do). Then you can get an archive node in a matter of a few hours and low I/O.

Three steps to get there:

1. Merge static files backend (WIP in Generalize cold DB: ColdStore trait + slot-keyed static archive dapplion/lighthouse#75)
2. Merge ERA files consumer support -> we get a fast block archive node (WIP in ERA file consumer and producer via LCLI #9273)
3. Merge ERB files consumer support -> we get a fast blob archive node