Flatten blobs to reduce allocations and copies

[flcl42]

Open questions:

• Should we change the code to use flat blobs?
• Should we flatten both v0 and v1 proofs and endpoints

Probably it should be a separate pr unless it will seem critical to do this optimization right in the main one

[benaadams]

Bunch of test errors?

  Failed Raw transaction in mempool form with accesslist, without to, non-trivial blobs [< 1 ms]
  Error Message:
   System.ArgumentException : Destination is too short. (Parameter 'destination')
  Stack Trace:
     at Nethermind.Serialization.Rlp.RlpStream.DecodeFlattenFixedByteArrays(Int32 itemLength) in /_/src/Nethermind/Nethermind.Serialization.Rlp/RlpStream.cs:line 1468