Normalize rqa epsilon by looking angle or illuminated area

[felixcremer]

I discussed with @MarkusZehner and he suggested to normalize the threshold we use for the detection of two time steps are similar by the looking angle or the illuminated area.
This is because the looking angle influences the variance of the time series by determining how deep the slice of the vegetation is, that is visible.

I opened a EODC support ticket to ask for this auxcillary information.
Another possibility would be to normalize by the time series variance, but that would mean, that we would have to touch the whole time series before computing the RQA TREND.