Add sepolia configs

[parithosh]

This PR adds the config files required by CL clients to instrument genesis. A summary of the chain configs are below:

• Genesis time: 1655733600
• Genesis fork version: 0x90000069
• Genesis validator count: 1550
• ALTAIR_FORK_EPOCH: 50
• BELLATRIX_FORK_EPOCH: 100
• TERMINAL_TOTAL_DIFFICULTY: 100000000000000000000000 (placeholder)
• DEPOSIT_CONTRACT_ADDRESS: 0x7f02C3E3c98b133055B8B348B2Ac625669Ed295D (Token gated deposit contract)

Values such as ETH1_FOLLOW_DISTANCE were left unchanged compared to other testnets.

Once the PR has been merged, the client teams can be expected to receive mnemonics for their genesis validators. It'd be great if once the client teams have setup their bootnodes, if we could increase the bootnode list.

[parithosh]

A widely known issue for testnets is that supply tends to eventually get short, Bots tend to drain faucets quite fast and eventually it gets harder to get test ether (Especially over years).

With the beaconchain genesis having embedded deposits, we have a unique opportunity to inflate the supply of sepolia eth once withdrawals are enabled. We can do this by changing the deposit balance value of the validator from 32 eth (Currently what the eth2-testnet-genesis tool sets) to a much higher number (1M eth). Once withdrawals are enabled, we can withdraw a small portion of the validator set to inflate supply over time as and when needed.

In order to avoid overflow issues, it was decided to keep the supply inflation to a reasonable factor.

In terms of the actual plan, we'd have 3 levers to then control on Sepolia:

• we assign 1M as the deposit value per validator (times 1550 validators), so the overall supply on Sepolia goes from ~320M to like ~1,850M (once withdrawals are enabled)
• we assign an additional 20 validators at genesis with a new mnemonic, their sole purpose is to be withdrawn at some point and inflate the supply by 20M without affecting the validator set negatively. EF Devops can hold the mnemonic for this and release as/when needed.
• Since we're using the token gated deposit contract, we already have the ability to onboard another 1,000,000,000 validators, each will show up as a 32 eth deposit (essentially 32x leverage). So we can additionally inflate the supply up to ~32,000,000,000 eth - as and when needed. This would be a last ditch effort that we'd avoid using unless we're in a dire situation.

I think a combination of the above 3 levers should mean we never have sepolia supply issues for years/decades to come(post withdrawals of course).

I've already created an alternate genesis state with the updated balances, updated in this branch but I can revert it if teams decide its not worth the risk or any other reason. Looking for feedback on the idea and if we've overlooked something (esp. regarding genesis state balance processing or how withdrawals work.

An argument against doing any of this is the token deposit contract, since it gives us the ability to inflate by ~32,000,000,000 eth we could just use that as the only lever. This would rely heavily on churning validators through the deposit/withdrawal queue though.

Since this would affect EL and CL teams + withdrawals, I'd be tagging @ajsutton @mkalinin @terencechain @tersec @paulhauner @dapplion @ralexstokes @lightclient @q9f @mcdee @MarekM25 @SuburbanDad @karalabe

[dapplion]

Lodestar can handle a max individual balance of 9007199254740991 due to internal types optimization; 1M ETH fits.
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER

[tersec]

Nimbus has the full range of uint64 available for balances in Gwei.

My only concern would be whether this affects the realism of the merge testing, but there are other testnets, so the tradeoff seems reasonable.

[ajsutton]

Great idea. This isn't meant to be a like MainNet beacon chain - it's a stable service we're providing to dapps for their testing primarily and ensuring good supply is important for that.

[ajsutton]

Btw, if this causes issues for any clients we should make sure we have reference tests and get the clients fixed. Once withdrawals are enabled you could cause a bunch of chaos of free by depositing a heap of ETH into your validator and then getting it back via partial withdrawals. I suspect no-one will have 9007199254740991 ETH they can spare for a few weeks (does that much ETH even exist?) but large but not unreasonable balances are certainly possible and the ETH used wouldn't be lost so we need to be sure we can handle it.

I'd be quite surprised if there are any problem though.

[mkalinin]

You want to be sure that BLS withdrawal credentials are used for these validators. Excess balance of those that are using eth1 credentials will be partially withdrawn in a few epochs after withdrawals feature gets enabled.

[ralexstokes]

high-level sounds good to me, I support taking advantage of all the various supply inflation levers we have

I left a few minor comments

[mcdee]

Lodestar can handle a max individual balance of 9007199254740991 due to internal types optimization; 1M ETH fits. https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER

@dapplion is that value in gwei? If so, it maps to around 9 million ETH and it's entirely possible to create a validator with that balance as Adrian mentions (certainly on testnets, anyway).

[protolambda]

Can the same nonce value that was used on Sepolia to deploy the smart contract still be used on mainnet? Ideally we deploy something to mainnet that blocks transfers into the contract, or allows an admin to drain any ETH sent to it by mistake.

[dapplion]

Lodestar can handle a max individual balance of 9007199254740991 due to internal types optimization; 1M ETH fits. https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/MAX_SAFE_INTEGER

@dapplion is that value in gwei? If so, it maps to around 9 million ETH and it's entirely possible to create a validator with that balance as Adrian mentions (certainly on testnets, anyway).

If a mega-whale of test ETH is bored and wants to kick Lodestar out of a testnet for the lolz then yes, it's entirely possible.

For the record: using Number instead of bigint for balances speeds up epoch processing very significantly in Mainnet, so it's worth the risk IMO.