The improved tx batching algorithm needs to track the rlp encoded byte
size of added blocks and the frame number, so they are now returned by
ChannelOut to avoid state and computation duplication.