FTR: Measure Server-Client Latency

[proof88]

Although GNS measures latency between server and clients, it is just the low level latency.
Higher level stuff such as PGE's incoming packet buffer, rendering delay, etc. is not calculated, but it would be interesting to see the overall latency between high level layer of server and client.
So a new message should be introduced which is periodically exchanged between server and client:

• tS0: timestamp of server sending out the message to client;
• tC0: timestamp of client receiving that message from server and when client sends the answer at high level;
• tS1: timestamp of server receiving message from client, and responding back to client (it also sends back value of tS1-tS0);
• tC1: timestamp of client receiving the 2nd message from server, for it it sends back value of tC1-tC0 to server.

This way server can display the higher level latency between server and client which is tS1 - tS0, and client's calculation tC1 - tC0.
Client can also display the latency between client and server which is tC1 - tC0, and server's calculation tS1 - tS0.

Not sure on the long run how this will be useful but definitely adds some value for debugging and/or benchmarking purpose in the future.