Previously, we would not care if the handshake timer timed out because
the server was too slow, or because we were too slow, this caused
connections to fail when the system was under heavy load.
This patch fixes this behaviour (and closes#2843) by checking if the
timeout delay was within margin of error of the max timeout.
The "ready to write" notifier we set up in generic socket connection is
really only meant to detect a successful connection. Once we have a TCP
connection, for example, it will fire on every event loop iteration.
This was causing IRC Client to max out the CPU by getting this no-op
notifier callback over and over.
Since this was only used by TLSv12, I changed that code to create its
own notifier instead. It might be possible to improve TLS performance
by only processing writes when actually needed, but I didn't look very
closely at that for this patch. :^)
Contrary to popular belief, not every implementation of TLS follows the
specs.
Some of them just drop the connection without sending a proper
close_notify, and we should handle that gracefully.
