Draft: Quantile filtering

[YBachmann]

Currently the minimum value of each column in the depthimage was used to calculate the distance values.

When dealing with noisy data it is benefitial to use e.g. the 10% quantile instead of the minimum value. This way we still get accurate data but ignore noise/outliers.

The parameter quantile_value controls which quantile should be used.

• Setting it to 0.0 will have the effect of using the minimum value of each column (the current behavior and default value).
• Setting it to 0.5 will use the median of each column.
• I got quite good results with using a quantile_value of around 0.1. This way a lot of noise/outliers was removed, but the laserscan data still contained obstacles, even if they only covered a small part (>10%) of the scan_height.

[YBachmann]

Well, seems I did something stupid... I tried to retroactively change the author data (name/email) of my commits because I used the wrong account for the commits. It seems that that did not got as planned. I will create a fresh merge request and do it the proper way.