Describe the bug
When sending sequential transactions on the RISE testnet from a colocated instance, transactions occasionally fail due to nonce errors. The transaction is added to the mempool but not processed, with an error indicating a "missing nonce". It appears that the local nonce tracking is ahead of the network's pending nonce.

To Reproduce
Steps to reproduce the behavior:

1. Deploy a script or app using eth_sendRawTransactionSync to send sequential transactions from a colocated EC2 instance.
2. Rapidly submit transactions, use spam bin here to reproduce.
3. Occasionally observe failure in processing due to nonce mismatch.
4. Error returned:

   Error: RPC error: (code: -32003, message: The transaction was added to the mempool but wasn't processed due to a missing nonce. Please submit a transaction with nonce 16706 first...
   

Expected behavior
Transactions should be accepted and processed in nonce order without skipping or failing due to an outdated pending_nonce value, assuming proper sequencing client-side.

Screenshots

Additional context

• The issue only occurs intermittently, often under load or during burst transaction sequences.
• Environment: EC2 colocated instance
• Network: RISE testnet
• Tooling: rust-web3-utils
• The transaction with nonce 16839 fails because the pending nonce on the node is still at 16706. Suprisingly big gap here