simx86: Do not stretch time if the EMUtime goes faster than the TSC.
The negative adjustments caused wrong multiplications and random hangs.
Now it looks like the random hangs are gone. There are only consistent
hangs left, all in DPMI it seems.


git-svn-id: https://dosemu.svn.sourceforge.net/svnroot/dosemu/trunk@1305 7758b857-130e-0410-ab9c-d8735a528d18