Comment 20 for bug 1861612

I happened upon a difference between the float to int calculations of epicsThreadSleep() vs epicsEventWaitWithTimeout() for WIN32.

epicsThreadSleep() has

> seconds *= mSecPerSec;
> seconds += 0.99999999; /* 8 9s here is optimal */

Perhaps Andrew can comment on why 8 9s is optimal.
(cf. 32f556637de8c0b9de626290f0859fc24f7ccf89)

epicsEventWaitWithTimeout() has

> tmo = ( DWORD ) ( ( timeOut * mSecPerSec ) + 0.5 );