Improve dangerzone calculation performance

- Right now, every time a single move has been made, either in a real move or a simulated move (all moves are simulated before being presented to the player), all pieces dangersquares are being recalculated.

Changing this approach to only update possibly affected pieces, should improve performance.

- Performance analysis results of bots, shows that UpdateDangerzones() is a major hot path. Any improvements should improve calculation speed of bots.

Start by locating UpdateDangerzones() and tracing how it is called after real and simulated moves, especially during bot analysis. Profile the current calculation path, then verify that only potentially affected pieces are updated and that bot move analysis completes faster without changing danger-square results.

Written by the indexing model from the issue text.