Eliminate the spawn_blocking hop in the multiproof pipeline.

• Pass the multiproof task's Sender (or a dedicated result sender) directly into the proof workers.
• Instead of recving here:

  reth/crates/trie/parallel/src/proof.rs

  Lines 126 to 130 in 792b82d

  	let proof_result = receiver.recv().map_err(|_| {
  	ParallelStateRootError::StorageRoot(StorageRootError::Database(DatabaseError::Other(
  	format!("channel closed for {hashed_address}"),
  	)))
  	})?;

  , we could forward a sender to the proofworker then we don’t need to spawn just to recv and send back the result
• Wire the worker-side result path (

  reth/crates/trie/parallel/src/proof_task.rs

  Lines 77 to 78 in 792b82d

  	/// Channel to send result back to original caller
  	result_sender: Sender<StorageProofResult>,

  ) directly into the multiproof task's inbox (

  reth/crates/engine/tree/src/tree/payload_processor/multiproof.rs

  Line 994 in 792b82d

  match self.rx.recv() {

  ).
• We would need a separate channel with a crossbeam select! because we cant use
  https://github.com/paradigmxyz/reth/blob/792b82d8956d91255c022abf7ea39a81781d0796/crates/engine/tree/src/tree/payload_processor/multiproof.rs#L[…]66
  due to different crates

Current:

MultiproofManager --(spawn_blocking)--> worker_rx.recv()
    └── result -> MultiProofTask channel

Proposed:

MultiproofManager -input-> worker
    worker -direct send-> MultiProofTask channel

Impact: Remove the spawn → recv → send roundtrip for both account and storage proof workers.

Additional context

https://docs.rs/crossbeam/latest/crossbeam/macro.select.html