test1.tgz and mlton rpm

[Matthew Fluet]

> > > When I compile test1 -native-optimize 2 and run it it appears to
stick
> > > in an infinite loop.
> >
> > conv2 of the regressions appears to exhibit similar behavior.
>
> and harmonic and smith-normal-form.  Clearly some interaction with
> IntInf.

I'll take a look at it later this evening.

I haven't tested anything with -native-optimize 2 in a long time; it's
somewhat left over from experiments that I was running with repeatedly
applying some of the native x86 peephole-optimizations.  Bottom line, it
never really helped appreciably, so I've never kept up with it.