The main motivation for importing yet another library (thankfully
it's not that big; 17 kB pre-gzip should be livable) is to get
accurate FENs for history support later. However, right now we
grab the following two nice side effects:
- We don't have to maintain our own castling and en passant logic
anymore.
- We don't have to send both UCI and prettyprinted moves over,
reducing the JSON size by something like 40% (I haven't checked
post-gzip, though). Currently we just send the UCI, but maybe
it should be reversed; we need to check if this causes problems
for CPU usage or something during normal updates, and which one
compresses better..