I wanted to add a series of quick solution sketches to my problems without one. These aren't technically full editorials, but I hope that they are enough for you if you are stuck. :-)

The first two subtasks were meant to reward various brute force approaches.

The third subtask was solvable by considering the ~x~ and ~y~ axes separately. Consider the ~x~ axis: if we let ~f(x)~ be the total payment of moving all chairs to position ~x~, we note that ~x~ is convex. Thus, we can binary search for the best position ~x~ (and repeat for the ~y~ axis).

The final subtask extends on the third subtask a bit, noting that the position where ~f(x)~ is maximized is either ~0~ or ~M~, allowing us to solve the problem in linear time. As ~N~ is very large, be careful with implementation.