Package has major issues with degraded Sockets

[Brondahl]

If the remote server is rebooted and we don't send a message within the 60s FIN packet timeout, then the TcpClient doesn't notice and keeps blindly sending logs into the void for ~18minutes, entirely dropping those logs.

Enabling KeepAlive or logging frequently both appear to mostly resolve it, but still leads to 1-2 log records getting lost when the destination reboots.

I've pulled this codebase into my project to refactor and fix it. Might raise a fixing PR eventually, but wanted to make sure this was documented on GH.