reg_genmove command

[ujh]

Simulates a genmove but doesn't play the resulting move. Respects the time settings but doesn't update the remaining time.

Closes #285

[coveralls]

Coverage increased (+1.7%) to 93.378% when pulling 58b7deb on reg_genmove into 110d8e5 on master.

[iopq]

I'm still incredulous that we have a test for the actual output string that we have to modify every time we add or remove a command.

[ujh]

That is not ideal, but then this whole GTP parsing code is a bit of a hack as we duplicate the list of commands. Once for list_commands and known_command and once for the actual command (a big match expression). That's why I think that test is still useful. Otherwise it would be possible to implement reg_genmove in the match expression but it won't appear in list_commands and so external tools wouldn't know it's there.

[iopq]

I remember when I had solved this issue 😉

[ujh]

In a more dynamic language I would have switched to a hash for this long ago. Use the keys for list_commands and known_command and execute the functions stored as values for the actual commands. Then everything would be in one place. I'm sure this is possible in Rust as well, but I haven't had the time (or the inclination) to figure that out yet.