eth_sendRawTransaction occassionally fails with a context cancelled in local devnet environment

[sieniven]

Issue

When running the local devnet, the eth api RPC layer seem to have some issues handling high transaction load during stress test environment where there are some issues with broadcasting multiple batches of transactions. It seems to take very long to submit a transaction (resulting in a context cancelled), and on the client layer, most clients will go into a retry logic to try to rebroadcast the txs. However, these txs are already known in the pool, resulting in failed broadcast.