core: parallelise nonce checking when processing blocks

ChainManager now uses a parallel approach to block processing where all
nonces are checked seperatly from the block processing process. This
speeds up the process by about 3 times on my i7