No failover to the next server in the MX list

[smancke]

With the change introduced in #4 no no failover to the next server in the MX list is done in case of a failure. To solve this, there are two questions:

• How should the timeout from the context be handled?
  (I guess, tho only solution is to spend the full timeout on the first server and have no failover in the case of a timeout)
• What to return in the case, multiple errors occur on different servers?
  (I guess, the most simple solutions would be to return either the first or the last error)