Optimizations

[GunshipPenguin]

Implement a variety of small optimizations (and a few small enhancements to existing features).

• Remove incrementally updated bitboards from Board class that could be calculated on the fly quicker
• Take advantage of branch prediction by surrounding code that only needs to be executed from time to time in conditionals
• Add Board::_movePiece() and use it instead of _removePiece() + addPiece() to save a few CPU instructions
• Bulk count leaves in perft calculations
• Don't use C++ exceptions, instead add a new [[ noreturn ]] fatal() function and compile with -fno-exceptions
• Improve showbb.py script to accept Python expressions rather than only single hex/decimal numbers
• Improve output of printboard and printmoves

Together, these optimizations improve raw perft speed by roughly 10%.