Retry bootstrap more than once

[iphydf]

Currently, bootstrap does:

• c-toxcore/toxcore/tox.c

  Line 761 in 43fc237

  dht_bootstrap(tox->m->dht, &root[i], public_key);

• c-toxcore/toxcore/DHT.c

  Line 1855 in 43fc237

  dht_getnodes(dht, ip_port, public_key, dht->self_public_key);

• c-toxcore/toxcore/DHT.c

  Line 1359 in 43fc237

  if (sendpacket(dht->net, ip_port, data, len) > 0) {

• c-toxcore/toxcore/network.c

  Line 642 in 43fc237

  return send_packet(net, ip_port, packet);

• c-toxcore/toxcore/network.c

  Line 624 in 43fc237

  const long res = sendto(net->sock.socket, (const char *)packet.data, packet.length, 0,

The only data we store about the bootstrap call is the ping array entry. If that one and only sendpacket call fails or the packet is dropped, we never try again.

Instead, we should store the fact that we just tried to bootstrap, add this bootstrap node as a friend, and ping it until it gets thrown out the k-buckets.

[emdee-is]

From: #2469 (comment)

PS: Dear Santa: A version of add_tcp_relay that loops through a set of tcp nodes and loops until dht_connected is true is very high on my wish list.

What I find I have to do in my testrunner for Tor testing is to integrate add_tcp_relay loops of a random subset the tcpnodes into the testrunner after a check of if we are connected, as the connection may drop. It's a big problem running over Tor that you would not see in clearnet testing,