Abandoned is a grid-based turn-based game. What could be easier?

Years ago I did such games the following way: * I have 2d array, every array item describes what is inside - wall, hero, enemy, etc. * When I need to check movements - I lookup 2d array * Need to check sight - lookup 2d array * Need to find path - lookup 2d array.

Now I get modern engine (Godot, but Unity has same problems) and visually put all my objects on scene instead of filling arrays in the code. That is amazing, really. But it becomes harder to check collisions the old good way (as I don't have grid coordinates, just "world" ones), so I have to use physics collisions. Not have to, but physics collisions are easier to use (on collide event - that's really helpful). So when I need to check movements - I use raycasts. When need to check objects on sight - use raycast or collision areas.

And that is easier to use with Godot and Unity, but for me - it is waste of CPU resources. 2d array would provide same functionality, and processing of that will not take time every frame to calculate collisions, and so on.

What do you think? Usage of physics is evil for grid-based games, or that is ok? Would you use physics for checkers or chess?