ethereum fdv is wrong

[serejandmyself]

[m1amgn]

#672

Fetch mainnet Ethereum supply from Etherscan V2 and independently validate it against CoinGecko, falling back safely when either provider fails or their values diverge. The Etherscan calculation follows the documented issuance, staking reward, and burned-fee components without subtracting beacon withdrawals.

Keep Tokenomics updates whole and failure-safe by refusing incomplete mainnet supply or Beacon snapshots, while preserving the existing Sepolia bonded-token behavior. Add focused provider and TVS tests, register them in the full suite, and document the optional server-side Etherscan key.

[serejandmyself]

@m1amgn still not showing on dev

[m1amgn]

#679

Ethereum FDV was stuck at $0. getTvs aborted the whole snapshot whenever the
Beacon request returned nothing — discarding the supply it had already fetched
successfully from Etherscan/CoinGecko — so total_supply was never written and
update-fdv kept skipping on its supply <= 0 guard.

The Beacon request itself always failed: the mainnet validator endpoint returns
~1 GB in one response and res.json() cannot exceed the V8 512 MB string limit.
Three retries burned 313s every hour and always ended empty. Server-side
filtering does not help — ?status=active and ?status=active_ongoing return the
full set byte for byte. Now folded incrementally with stream-json: 15 MB heap
over the full gigabyte, 107s to a real result.

Mainnet no longer indexes validators one by one — jobs/get-nodes.ts queries the
database per node, so ~890k active validators would mean ~890k round trips.
Aggregate stake is streamed instead. Testnets keep the per-node path.

Also renames the account Tokens tab to Movements (en/ru/pt) — it lists wallet
transfers, not holdings. Route and message namespace unchanged.