[MLton] interrupted system call

[Matthew Fluet]

> > It also doesn't seem to capture the property that I proposed, which was
> > that the signals were blocked until the signal handler got a chance to
> > run.
>
> I guess you've given up on this property by using ML to block the
> signals instead of using GC_handler?

Yes, because I realized it was overkill for "making progress".  If we are
inside a critical section, we want to finish the system call, so we block.
But, once we're done with the system call, we've "made progress", so if
other signals arrive before you leave the critical section, the programmer
should see them when control finally leaves the critical section.