Seeking other's benchmarks

[JeffreySarnoff]

@KlausC Everything is reworked; I am using one type rather than two for each FastRational.
There can be no clashes of expected types now. And, fortuitously, it does live up to its name.

Would you mind getting the current master and running the benchmarks file, then just cut and paste whatever it prints in this issue. It would be good to average some results rather than just use my computer's timings. Of course, let me know of anything else you may notice. Thank you.

[KlausC]

The reported speedup is really impressive for small integers. I added my versioninfo() for reference.
`
julia> print(relspeeds)
relative speeds
(32) (64)
mul: 30.6 28.3
muladd: 21.5 21.6
add: 17.4 17.4
poly: 14.2 27.2
matmul: 10.3 12.5
matlu: 2.1 2.2
matinv: 2.8 2.6

julia> versioninfo()
Julia Version 1.0.4
Commit 38e9fb7f80 (2019-05-16 03:38 UTC)
Platform Info:
OS: Linux (x86_64-redhat-linux)
CPU: Intel(R) Core(TM) i7-3610QM CPU @ 2.30GHz
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-6.0.0 (ORCJIT, ivybridge)
Environment:
JULIA_EDITOR = /usr/bin/vim -R -c"set nonumber"
`