feat: add new chain Ethereum Sepolia Testnet

[thesems]

Adds Ethereum Sepolia Testnet (ETHS, chainId 11155111) as a supported EVM chain.

Changes

• supportedChains.evm.ts — chain entry: ChainKey.ETHS, native ETH, mainnet: false, relayerSupported: false, explorer sepolia.etherscan.io, RPC ethereum-sepolia-rpc.publicnode.com.
• multicall.ts — canonical multicall3 0xcA11bde05977b3631167028862bE2a173976CA11.
• foundry.ts — foundryChainNameMap → sepolia (matches config/networks.json in contracts).
• coins.ts — native ETH, USDC (0x1c7D4B196Cb0C7B01d743Fbc6116a902379C7238), and WETH (0xfFf9976782d46CC05630D1f6eBAb18b2324d6B14, matching the chain's wrappedNativeAddress in contracts).

Also: Arbitrum Sepolia WETH correction

wrappedTokens[ChainId.ARBS] moves 0x2836ae2e… → 0x980B62Da83eFf3D4576C647993b0c1D7faf17c73.

The old address landed in #243 and never matched wrappedNativeAddress for arbitrumsepolia in the contracts repo's config/networks.json, so it isn't the WETH the deployed diamond wraps/unwraps with. Both are real WETH deployments on 421614, but the old one is effectively unused (~0.36 total supply vs ~3,630). Included here because the new ETHS entry above is added under exactly the rule ARBS was violating — the matching backend-side fix is in the paired BE PR.

Dependencies / merge order

Pinned to @lifi/types@18.3.0-alpha.0 (lifinance/types#569) for BE validation. Do not merge yet — per the agreed sequence:

1. types alpha ✅ 18.3.0-alpha.0 published
2. use it in data-types ✅ this PR
3. publish data-types alpha
4. validate both on BE (pnpm check failures expected until tenderly-sim is bumped)
5. revert the alpha pins in types + data-types, merge both, publish stable
6. bump tenderly sim
7. update BE with all three packages

So the @lifi/types pin here still needs to move to stable 18.3.0 before this merges.