If it returns 0 that just means we hit the timeout. I suppose we could skip
the timer checks if there are a non-zero number of fds marked as well, but
I'm not sure that's a great idea since it will add some latency.
Also run it across the whole tree to get everything using the One True Style.
We don't yet run this in an automated fashion as it's a little slow, but
there is a snippet to do so in makeall.sh.
Move the bulk of exec() into a new pump(). Since SDL wants to drive the
event loop itself, this is a requirement. We also add a WaitMode flag to
allow for immediately pumping events -- again, this is required because
SDL wants to be in full control of the event loop, and not let us wait.
The scheduler expects m_select_timeout to act as a deadline. That is, it
should contain the time that a task should wake at -- but we were
directly copying the time from userspace, which meant that it always
returned virtually immediately.
At the same time, fix CEventLoop to not rely on the broken select behavior
by subtracting the current time from the time of the nearest timer.
