GitLab

This is what we recommend people to use for best performance, so we
should use it in our benchmarks.

The benchmark is not enabled by default as the runstdtest.prl script
doesn't find the automated stdout files on the first run, thus requiring
make to be run twice.

Newer versions (corresponding to what's currently on the shootout page)
have been added under nofib/shootout in an attempt to make these easier
to run.

I haven't checked if the output is right or not, but it looks roughly
consistent with the existing variety of values.

The benchmark implements the Kahan summation algorithm and tests unboxed
arrays and floating point arithmetic in tight loops.

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>