add native-is-erc20 chain smoke test variants

[david-uniswap]

Summary

Stacked on top of #139. Adds V2/V3/V4 smoke test variants in script/smoke/native-is-erc20/ for chains where the native gas token is itself an ERC-20 (Arc, CELO, and similar Circle-style partner chains).

Motivation

On these chains the WETH9 dependency in the deploy gets wired to UnsupportedProtocol (or any other placeholder that reverts on deposit/withdraw), because there is no real IWETH9-compliant wrapper. The standard smoke tests from #139 all halt at IWETH9(weth).deposit{value:...}() on these chains, even though the contract deployment itself is functional.

These variants drop the WETH leg entirely and use two freshly-deployed TestTokens to exercise the same v2/v3/v4 + UR code paths.

What's in this PR

File	Tests
script/smoke/native-is-erc20/V2SmokeNativeIsERC20.s.sol	Pair creation, addLiquidity, swapExactTokensForTokens via Router02
script/smoke/native-is-erc20/V3SmokeNativeIsERC20.s.sol	createPool + initialize, NPM mint, SwapRouter02 exactInputSingle
script/smoke/native-is-erc20/V4SmokeNativeIsERC20.s.sol	initializePool, Permit2 dual-approval, modifyLiquidities mint, StateView reads, UR V4_SWAP command

Chain-agnostic via vm.parseJsonAddress reads from deployments/json/<chainId>.json, matching #139's pattern.

Tested on Arc Mainnet (chain 5042)

All three variants ran end-to-end on Arc with the WETH9-as-UnsupportedProtocol deploy from #144:

• V2: pair created, 1000:1000 liquidity, swap 1 A -> 0.996 B (the standard 0.3% fee math)
• V3: pool initialized at sqrtPrice 1:1, position Add v2 to packages #1 minted with 1000e18 liquidity, exactInputSingle 1 A -> 0.996 B
• V4: pool initialized, Permit2 allowances granted to PositionManager + UR, position minted via modifyLiquidities(MINT_POSITION + SETTLE_PAIR), UR V4_SWAP command swaps A -> B successfully

When to use

• Standard chains (mainnet, OP-stack L2s, Arbitrum, etc.) -> use the regular smoke tests in script/smoke/ from add chain-agnostic smoke tests for v2, v3, v4 (with UR swap) #139.
• Chains with native-is-ERC20 + non-IWETH9 wrapper -> use the native-is-erc20/ variants in this PR.

Note on gas estimation

On Arc specifically, forge's eth_estimateGas returns tight per-tx estimates that can cause the v2/v3 swap txs to OOG during the K-invariant check. Workaround: pass --gas-estimate-multiplier 200 when broadcasting. Likely worth surfacing in #139's README too once both PRs land.

Test plan

• Compile cleanly
• V2 variant: end-to-end on Arc (tx hash 0xdadd632ab0db9ddb43148b93c7b9b01f725fe22800403d0622403e15ac283d3c is the final swap)
• V3 variant: end-to-end on Arc
• V4 variant: end-to-end on Arc

🤖 Generated with Claude Code