[cpt_deciation_analysis] update-filters for multi-scan deviations

[hermannsblum]

See the todo for updateAveragePlaneDeviation. Updates to the deviation scores should be proper filter updates with some exponential decay.

[jstiefel]

Thinking about this, new results should be weighted depending on how far away from the LiDAR the walls are and how sharp the incident angle is. Not with an exponential decay since results do not change w.r.t. time.