Added best node search as an alternative to PVS

[redactedontop]

Howdy @lithander!

Best node search (or BNS - as drafted in https://www.bjmc.lu.lv/fileadmin/user_upload/lu_portal/projekti/bjmc/Contents/770_7.pdf), is an alternative to both MTD(f) and PVS with much faster convergence (20% faster than MTD(f) in the best case and still 4% faster in the worst). And, on top of that, it's still stable with "bad" first guesses, because of how it operates.
No mainstream engine uses it, would love to see how it operates with Leorik.

Thanks,
Alex <3
PS: I also did a couple of style changes and comment spell checking - if you don't like those, I'll remove them for ya

[redactedontop]

Hey!

I've found some mistakes in my code so far (I think they're fixed). Before merging, can you do a full review? I don't wanna risk anything.

Thanks,
Alex <3
PS: My contribution was adapted from https://www.bjmc.lu.lv/fileadmin/user_upload/lu_portal/projekti/bjmc/Contents/770_7.pdf, so that's a good fact-checking guide.

[redactedontop]

I think there's also some redundancy with the final full-window search. If no candidates are found, that means 2 of the found moves are equal, so you can pick any of these (or do some more sophisticated picking logic)

[tissatussa]

BNS seems cool ! I quickly looked into the PDF.

[redactedontop]

Hey @tissatussa!

Thanks for acknowledging me. I have both Redacted Leorik (don't judge 😭) and Leorik playing games while I watch. It seems like mine is a bit weaker, which doesn't make sense: the only explanation is that I fucked up something in the code when compared to PVS.
Do you have the free time to do a code review?

Thanks,
Alex <3

[redactedontop]

Welp, update: my version is losing most games. Gonna look into that in a second.
Gotta study xD.

[tissatussa]

Do you have the free time to do a code review?

sorry, no .. i'm a little into Python but no C#
i can do tests in CuteChess though, letting engines play games and analyse those.