[Cid-oe]

Closes #130

[ShisatoYano]

Hi @Cid-oe, thank you for picking up #130! The PR covers every point of the issue in a small, focused diff: the speed limiting now lives in one _limit_speed() with the original order kept, input is renamed to motion_input (docstring included), update_by_localizer() has its docstring, and the history plot is updated with set_data instead of being re-created. I checked it locally: all 96 tests pass, and State.draw cumulative time in pure_pursuit_path_tracking drops from 0.27 s to 0.12 s over 251 frames.

Please address the following before merge:

1. The trajectory is now drawn under the course. Because the history line is created once and kept, the course line (re-created every frame) is drawn on top of it, and the black tracking-error dots are mostly hidden. On the final frame of pure_pursuit_path_tracking, 5,310 px differ from main. Adding zorder=2.5 to the axes.plot(...) call brings it back to main's look (11 px difference).
2. Please add a short comment explaining that hist_plot is intentionally not appended to elems (it is kept so set_data can update it). Every other draw() appends to elems, so readers may "fix" it otherwise.
3. Small style points: rename it to _hist_plot since it is internal, and add a one-line docstring to _limit_speed().
4. (Optional) To stay safe if the same State is ever drawn on a new axes, you could re-create the line when self._hist_plot.axes is not axes.

Thanks again, looking forward to the update!