[jmg-duarte]

Description

Follow up to the discussion in #5114 (comment), this PR introduces the TxReceiver trait, which "covers" the single mempools (and in the future, the single builders) as a single "transaction receiver".

Changes

• Transactions are signed before being passed to the Receiver layer (old Mempool)
• Single receivers are abstracted over using TxReceiver (the generic parameter was done on purpose to ensure the collection of receivers only carries a single type!)
• Adds a NonceSource trait for "nonce-enabled transaction receivers" — e.g. nodes can returns nonces (TxReceiver + NonceSource) but builders can't (TxReceiver)

How to test

Existing tests

[claude]

Claude finished @jmg-duarte's task in 4m 17s —— View job

---

Review: Abstract over single transaction receivers

Nice refactor — pulling signing out of the submission transport and hiding single mempools/relays behind TxReceiver reads cleanly, and the generic Mempools<R> keeping the collection monomorphic is a nice touch. I verified the metric/label strings (Mempool(name)), the pending_gas_price replacement semantics, chain-id handling, and the account set registered into Wallet are all preserved from the old flow.

A couple of non-blocking notes, posted inline:

• wallet.rs — the new doc comment claims the tx is signed once and the same bytes sent to every receiver, but signing actually happens per-receiver inside submit (and final_gas_price differs per receiver), so the shared-bytes optimization isn't realized here. Not a regression, just a comment that overstates what this PR does.
• mempools.rs — nonces are now read from the shared self.ethereum node instead of each receiver's URL (the relay nonce-via-proxy comment was removed). Looks intentional and correct; flagging for confirmation.

Everything else looks good. I couldn't run cargo check/tests in this environment (command not permitted), so compilation relies on your local run / CI.

[MartinquaXD]

Wouldn't this interface cause issues for block builders?
Or are you planning to create a temporary struct which knows the block range the tx should be submitted at such that you don't have to pass it here anymore?