Make Move Ordered

[isaacl]

Use a natural ordering on Move rather than an external comparator. This improves readability, but also jmh benchmarks by about 10%.

[niklasf]

I don't see the benchmark improvement at all (running with lots of forks to deal with the bimodal distribution). But it is more readable ...

[isaacl]

jmh on HuffmanPgnBench.decode is oddly bimodal. During a fork's runs, the times are consistent, but different forks/runs are not consistent.

$ sbt 'project benchmarks; jmh:run -wi 1 -i 3 -f10 HuffmanPgnBench.decode'

before PR:

• 9 forks are 7.00ms/op
• 1 fork is 5.66ms/op

with PR:

• 7 forks are 6.75ms/op
• 3 forks are 4.98ms/op

So, on my machine, the 'good' runs are 10% faster locally, and the 'bad' runs are about 5% faster.

[isaacl]

But this is the second PR here where I saw improvements locally and the other person didn't... so idk.