Optimize discover protocols

[LilianBoulard]

Currently, when sending a discover request, the reply contains every single contact/node the contact knows.
In order to optimize the network communications, the author of the request should provide the list of the IDs of the contacts/nodes he knows.

Secondly, in order to avoid DOS attacks (or such) on the network, we should have control measures:

1. Verify the contact information specified in the request is the same that sends the actual request.
2. Verify that this contact did not send a similar request during the last x hours (24, 48, etc)
   2b. If the request has been made in the past, we can verify from our raw_requests database that the information is consistent.