Add binning and box masking.

[jgoppert]

Adds binning and box filter. The binning makes this compliant with slam_toolbox expectations. Instead of sending as many sample as are available, this sets a fixed number of angular bins and fills them based on the samples.

The box filter is similar to the angular mask, but can be used to mask out regions on your robot more precisely than angular masking. If you have a narrow mast for instance, box binning will allow you to ignore it based on x, y position instead of based on angle. This is more robust and you get to keep more of the lidar scan. We also tried using the box filter from ros2_laser_filters, but the cpu load was very high compared to this implementation.