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.