The `branch` of the operation contains enough information to induce
the `net_id`, and the code of the validator/prevalidator is now mature
enough to efficiently determine the `net_id` of an incoming operation.
The single validation module is split in multiple (simpler)
modules. In the process, we introduce one "validation worker" per
peer. This worker handle all the `New_head` and `New_branch`
advertised by a given peer. For so, it sends "fetching request" and
"validation request" to respectively the `Distributed_db` and and the
`Block_validator`. These two global workers are responsible of the
'fair' allocation of network and CPU ressources amongst the connected
'peers'.