Memleak when re-connecting a Jardin Client

[Aleksander Morgado]

When for example jardin_client_send_receive() gives a timeout, re-connection try happens automatically in the Jardin client, so it will actually disconnect() the connection module, and connect() the connection module again.

The problem is that when 2nd connect() in the same connection module happens, jardin_connection_module_socket() is called again, and both _host_address and _peer_address struct address-es are re-allocated again, without deallocating them previously.

The fix could be just:
 * If already allocated, just memset() them to 0

[Aleksander Morgado (aleksander-m)]

Fixed in trunk now.