Add robinhood chains to ethgas readers and ethrpc networks

[pkieltyka]

Adds robinhood (4663) and robinhood-testnet (46630) to the two chain-ID registries that treat Arbitrum chains specially. Both are Arbitrum Orbit chains running Arbitrum Nitro.

1. ethgas/readers.go — gas price paid reader

 	200:    arbitrumGasPricePaidReader, // arbitrum xdai
+	4663:   arbitrumGasPricePaidReader, // robinhood
+	46630:  arbitrumGasPricePaidReader, // robinhood testnet

Without an entry these chains fall through to DefaultGasPricePaidReader, which reports each transaction's GasFeeCap — the maximum the sender was willing to pay. On these chains that overstates what was actually paid:

• eth_maxPriorityFeePerGas returns 0 on both, and feeHistory p50 rewards are 0 across 100 blocks
• The docs state ordering is "first-come, first-served based on sequencer arrival time" and that "increasing your fee will not shift your transaction ahead of others already in the queue"

So tips cannot affect inclusion and effectively nothing is paid above base fee. arbitrumGasPricePaidReader reports block.BaseFee(), which is what was really paid. Affects gas statistics only, not correctness.

2. ethrpc/networks.go — Networks registry

+	4663: {
+		Name:                "robinhood",
+		ChainID:             4663,
+		NumBlocksToFinality: 50,
+	},
+	46630: {
+		Name:                "robinhood-testnet",
+		ChainID:             46630,
+		NumBlocksToFinality: 50,
+	},

ethreceipts consults this for NumBlocksToFinality when a caller doesn't set its own. 50 matches every arbitrum entry in the map (arbitrum, arbitrum-testnet, arbitrum-nova). Without an entry both chains fell back to DefaultNumBlocksToFinality of 40 — workable, but implicit and inconsistent with the family.

Worth flagging: 50 blocks is ~5s at robinhood's 100ms block time, against ~12.5s for arbitrum at 250ms. Reorg depth is block-based so matching the family's depth is the sensible reading, but the wall-clock window is shorter. Easy to raise if you'd prefer time parity.

There's no ArbitrumChain field on the Network struct — only OptimismChain — so nothing else to set.

Verification

go build ./... passes and gofmt is clean for both files. go test ./ethgas/... passes.

go test ./ethrpc/... fails with panic: ethtest: unable to connect to testchain — confirmed pre-existing by running the same test on an unmodified tree, where it panics identically. That suite needs a local testchain that isn't running in this environment.

Context

Companion to 0xsequence/relayer#189, which adds the same two chain IDs to the relayer's isArbitrum() — that one does affect correctness, since it was silently overriding MinGasTip to 1 Gwei on a ~0.05 Gwei chain. Chain configs are in 0xsequence/devops#4687.

Both repos now carry hardcoded chain-ID allowlists for this, so each new Orbit chain needs a release in both. Replacing that with a config flag is planned as a separate follow-up.

🤖 Generated with Claude Code